Timeline

[USER] You are a content analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate blog posts and articles for signal value. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, Supabase, Vercel, voice AI, agentic coding, Discord/Telegram bots, cron automation, Python, TypeScript. Return ONLY valid JSON: { "tldr": "one sentence summary of the post's core insight or claim", "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "must-read" | "worth-skimming" | "skip", "verdict_reason": "one sentence" } Verdict: - "must-read": direct actionable insight for Mike's active work - "worth-skimming": adjacent signal, may matter later - "skip": off-topic or generic Be OPINIONATED. Not neutral. Source: Simon Willison's Weblog Title: GDS weighs in on the NHS's decision to retreat from Open Source URL: https://simonwillison.net/2026/May/17/gds-weighs-in/#atom-everything Summary: <p><strong><a href="https://shkspr.mobi/blog/2026/05/gds-weighs-in-on-the-nhss-decision-to-retreat-from-open-source/">GDS weighs in on the NHS's decision to retreat from Open Source</a></strong></p> Terence Eden continues his coverage of the NHS' <a href="https://shkspr.mobi/blog/2026/05/nhs-goes-to-war-against-open-source/">poorly considered deci Analyze per your instructions. Return ONLY valid JSON, no markdown fences. [ASSISTANT] {"tldr":"UK's Government Digital Service publicly pushes back on the NHS's decision to retreat from open source software development.","why_mike_cares":"Low overlap — UK government policy debate, no agent/LLM/automation signal.","verdict":"skip","verdict_reason":"Off-topic policy commentary with no actionable insight for Mike's AI agency or builder work."}

[USER] hey add to vault obsddidian and skills to agents or reosurces etc also how can this help my set up [ASSISTANT] All wired. Here's the full picture: ## What got added this turn ### Obsidian vault enhancements (D:/Ecosystem/vaults/feedly-api-docs/) - **MOC.md** — Map of Content with section emojis, wikilinks, tag map, quick query patterns table - **README.md** — Vault usage guide, graph view setup, related projects - 77 content pages from before, untouched ### Skill — `feedly-api` - Source: `D:/ClaudeDev/00_GITHUB/merlino-skills/feedly-api/` (git committed + pushed) - Runtime: `C:/Users/mikem/.claude/skills/feedly-api/` (live now, visible in skill list) - Files: `SKILL.md` + 4 references (pagination, limits/errors, webhook-events, use-cases) - Triggers: Feedly, AI Feed, threat intel, IoC, CVE, malware, threat actor, cyberattacks, TTP, feedly webhook, feedly Ask AI ### Agent ownership — Raven - Added `feedly-api` to Raven's skills list - Added Feedly entry to Raven's Research APIs block - Cleaned up 3 duplicate `research-pipeline` entries while I was in there ## How this helps your setup — direct fit to your stack The Feedly API isn't another scraper. It's **curated + AI-summarized + webhook-delivered signal**. Concrete plug-ins to what you already run: | Your existing piece | Feedly add-on | What it unlocks | |---|---|---| | **Shakespeare** (content) | AI Feeds per vertical (plumbing, HVAC, electrical) → webhook to Discord `#content-radar` | Constant fresh industry news for blog/PAA topics, no manual feed-hunting | | **Einstein** (SEO) | Daily `/v3/ask-ai` query: *"5 most actionable trends this week"* per vertical | Auto-generated topic briefs for the next BMM piece | | **MERA / cold outbound** | `Get Company Metadata` + `Top Stories` per prospect domain | Personalized first-line per email, generated by API | | **Radar skill** (you have this) | Webhook handler pushes new Feedly entries into radar's queue | Single inbox for YouTube + GitHub + X + Feedly signals | | **GHL contacts** | Web Alert webhooks → GHL contact note when competitor mentioned in metro | Account managers get pinged hours after a new competitor opens | | **HawkeyePanel** | IoC lookup against caller voice transcripts for phishing kit domains | Catches social-engineering calls (high-leverage MSP/client play) | | **Master Brain** | Live MCP category `feedly` already added; restart CC to activate | Any session can semantically query the 509-chunk SOP for endpoint details | ## Strategic angle worth thinking about Feedly TI is **enterprise-priced**. If you already pay for it (or are evaluating), the wedge is this: most local-service marketing agencies cannot ship credible cyber-threat briefings or competitor-mention pipelines. With Feedly + your fleet + Shakespeare + Picasso, you can package a **"weekly threat brief"** or **"local market intelligence report"** as a productized $497–$1,997/mo add-on to MSPs, MSSPs, IT firms, and law firms (all higher-value than plumbers). That's a margin-stack on top of your existing pipeline. If you don't pay for Feedly TI, the skill still works for the general Feedly API (AI Feeds, Search, Ask AI), and the SOP site is a sales asset on its own. ## Next time you want to use it - Just ask Raven anything Feedly-related — the skill auto-triggers on the keywords. - Or hit the RAG directly: `curl -s -X POST https://feedly-api-docs.vercel.app/api/chat -H 'Content-Type: application/json' -d '{"message":"..."}'` - Or open Obsidian on `D:/Ecosystem/vaults/feedly-api-docs/` and start at `MOC.md`. **Reminder:** Restart Claude Code once when convenient — that's what activates the new `feedly` category in the merlino-sops MCP (it's already registered, just needs the server reload).

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: shadcn-ui/ui Stars: 114596 Language: TypeScript Topics: base-ui, components, laravel, nextjs, radix-ui, react, shadcn, tailwindcss, tanstack, ui, vite Description: A set of beautifully-designed, accessible components and a code distribution platform. Works with your favorite frameworks. Open Source. Open Code. README (first 3000 chars): # shadcn/ui A set of beautifully designed components that you can customize, extend, and build on. Start here then make it your own. Open Source. Open Code. **Use this to build your own component library**.  ## Documentation Visit https://ui.shadcn.com/docs to view the documentation. ## Contributing Please read the [contributing guide](/CONTRIBUTING.md). ## License Licensed under the [MIT license](./LICENSE.md). Analyze this repo per your instructions. Return ONLY valid JSON, no markdown fences. [ASSISTANT] { "what_it_is": "Open source library of accessible, customizable React components built on Radix and Tailwind that you copy into your codebase instead of installing as a dependency.", "stack": ["React", "TypeScript", "Tailwind CSS", "Radix UI", "Next.js"], "why_mike_cares": "ShadCN is part of Mike's default stack (Next.js + ShadCN + Tailwind + Supabase) and powers ClawControl, HawkeyePanel, BirdsEyeROI, and every SOP site he ships.", "verdict": "check-it-out", "verdict_reason": "Core dependency of nearly every Mike project, so staying current on upstream changes directly impacts active builds." }

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: livekit/python-sdks Stars: 356 Language: Python Topics: agents, ai, python, real-time, webrtc Description: LiveKit real-time and server SDKs for Python README (first 3000 chars): <!--BEGIN_BANNER_IMAGE--> <picture> <source media="(prefers-color-scheme: dark)" srcset="/.github/banner_dark.png"> <source media="(prefers-color-scheme: light)" srcset="/.github/banner_light.png"> <img style="width:100%;" alt="The LiveKit icon, the name of the repository and some sample code in the background." src="https://raw.githubusercontent.com/livekit/python-sdks/main/.github/banner_light.png"> </picture> <!--END_BANNER_IMAGE--> [](https://pypi.org/project/livekit/) [](https://pypi.org/project/livekit-api/) # 📹🎙️🐍 Python SDK for LiveKit <!--BEGIN_DESCRIPTION--> Use this SDK to add realtime video, audio and data features to your Python app. By connecting to <a href="https://livekit.io/">LiveKit</a> Cloud or a self-hosted server, you can quickly build applications such as multi-modal AI, live streaming, or video calls with just a few lines of code. <!--END_DESCRIPTION--> This repo contains two packages - [livekit](https://pypi.org/project/livekit/): Real-time SDK for connecting to LiveKit as a participant - [livekit-api](https://pypi.org/project/livekit-api/): Access token generation and server APIs ## Using Server API ```shell $ pip install livekit-api ``` ### Generating an access token ```python from livekit import api import os # will automatically use the LIVEKIT_API_KEY and LIVEKIT_API_SECRET env vars token = api.AccessToken() \ .with_identity("python-bot") \ .with_name("Python Bot") \ .with_grants(api.VideoGrants( room_join=True, room="my-room", )).to_jwt() ``` ### Creating a room RoomService uses asyncio and aiohttp to make API calls. It needs to be used with an event loop. ```python from livekit import api import asyncio async def main(): lkapi = api.LiveKitAPI("https://my-project.livekit.cloud") room_info = await lkapi.room.create_room( api.CreateRoomRequest(name="my-room"), ) print(room_info) results = await lkapi.room.list_rooms(api.ListRoomsRequest()) print(results) await lkapi.aclose() asyncio.run(main()) ``` ### Using other APIs Services can be accessed via the LiveKitAPI object. ```python lkapi = api.LiveKitAPI("https://my-project.livekit.cloud") # Room Service room_svc = lkapi.room # Egress Service egress_svc = lkapi.egress # Ingress Service ingress_svc = lkapi.ingress # Sip Service sip_svc = lkapi.sip # Agent Dispatch dispatch_svc = lkapi.agent_dispatch # Connector Service connector_svc = lkapi.connector ``` ## Using Real-time SDK ```shell $ pip install livekit ``` ### Connecting to a room see [room_example](examples/room_example.py) for full example ```python from livekit import rtc async def main(): room = rtc.Room() @room.on("participant_connected") de [ASSISTANT] {"what_it_is":"Official Python SDKs for LiveKit providing real-time WebRTC connectivity and server-side API access (rooms, egress, ingress, SIP, agent dispatch) for building voice/video AI apps.","stack":["Python","WebRTC","asyncio","aiohttp","LiveKit","SIP"],"why_mike_cares":"Direct overlap with voice AI work, SignalWire/telephony stack, and agentic real-time apps; LiveKit is the canonical infrastructure for Python voice agents and competes with/complements SignalWire for AI calling.","verdict":"check-it-out","verdict_reason":"Voice AI and real-time agent infra is a core active domain (Vox lead, call-sentiment-AI, HawkeyePanel) and LiveKit's Python SDK is the standard for building agent voice pipelines."}

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: CallFire/callfire-api-client-python Stars: 0 Language: Python Topics: api, call, callfire, messaging, mms, python, sms, text, voice Description: CallFire API v2 Python client README (first 3000 chars): # CallFire API v2 Python client See setup instructions at **[CallFire Developers Site](https://developers.callfire.com/callfire-api-client-python.html)** Analyze this repo per your instructions. Return ONLY valid JSON, no markdown fences. [ASSISTANT] {"what_it_is": "Official Python client for CallFire's v2 API for sending SMS, MMS, and voice calls.", "stack": ["Python", "CallFire API", "REST"], "why_mike_cares": "Low overlap — Mike uses CallRail for call analytics and SignalWire/GoHighLevel for telephony, not CallFire.", "verdict": "skip", "verdict_reason": "CallFire isn't in Mike's stack and the repo has zero stars, signaling minimal community traction."}

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: pydantic/pydantic Stars: 27793 Language: Python Topics: hints, json-schema, parsing, pydantic, python, python310, python311, python312, python313, python39, validation Description: Data validation using Python type hints README (first 3000 chars): # Pydantic Validation [](https://github.com/pydantic/pydantic/actions?query=event%3Apush+branch%3Amain+workflow%3ACI) [](https://coverage-badge.samuelcolvin.workers.dev/redirect/pydantic/pydantic) [](https://pypi.python.org/pypi/pydantic) [](https://anaconda.org/conda-forge/pydantic) [](https://pepy.tech/project/pydantic) [](https://github.com/pydantic/pydantic) [](https://github.com/pydantic/pydantic/blob/main/LICENSE) [](https://docs.pydantic.dev/latest/contributing/#badges) [](https://docs.pydantic.dev/latest/llms.txt) Data validation using Python type hints. Fast and extensible, Pydantic plays nicely with your linters/IDE/brain. Define how data should be in pure, canonical Python 3.10+; validate it with Pydantic. ## Pydantic Logfire :fire: We've launched Pydantic Logfire to help you monitor your applications. [Learn more](https://pydantic.dev/logfire/?utm_source=pydantic_validation) ## Pydantic V1.10 vs. V2 Pydantic V2 is a ground-up rewrite that offers many new features, performance improvements, and some breaking changes compared to Pydantic V1. If you're using Pydantic V1 you may want to look at the [pydantic V1.10 Documentation](https://docs.pydantic.dev/) or, [`1.10.X-fixes` git branch](https://github.com/pydantic/pydantic/tree/1.10.X-fixes). Pydantic V2 also ships with the latest version of Pydantic V1 built in so that you can incrementally upgrade your code base and projects: `from pydantic import v1 as pydantic_v1`. ## Help See [documentation](https://docs.pydantic.dev/) for more details. ## Installation Install using `pip install -U pydantic` or `conda install pydantic -c conda-forge`. For more installation options to make Pydantic even faster, see the [Install](https://docs.pydantic.dev/install/) section in the documentation. ## A Simple Example ```python from datetime import datetime from typing import Optional from pydantic import BaseModel class User(BaseModel): id: int name: str = 'John Doe' signup_ts: Optional[datetime] = None friends: list[int] = [] external_data = {'id': '123', 'signup_ts': '2017-06-01 12:22', 'friends': [1, ' [ASSISTANT] {"what_it_is":"Python library for data validation and parsing using type hints, with strict schemas, JSON serialization, and IDE-friendly models.","stack":["Python","Type Hints","JSON Schema","Rust (pydantic-core)"],"why_mike_cares":"Pydantic is the backbone of LLM tool-calling, structured outputs, MCP servers, and Python agent frameworks Mike uses daily (Claude Agent SDK, FastAPI automations, SignalWire/GHL pipelines).","verdict":"check-it-out","verdict_reason":"Foundational dependency across Mike's Python agent stack and any structured-output LLM work he ships."}

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: bmad-code-org/bmad-module-creative-intelligence-suite Stars: 81 Language: JavaScript Topics: Description: A BMad Method Compliant stand along module that has agents and workflows to help bring out the creativity of the user through various exercises and disciplines. More will come over time - this was meant as a bmad module tech demo README (first 3000 chars): # Creative Intelligence Suite [](https://www.npmjs.com/package/bmad-creative-intelligence-suite) [](LICENSE) [](https://www.python.org) [](https://docs.astral.sh/uv/) [](https://discord.gg/gk8jAdXWmj) **Think differently.** A collection of agents and workflows for innovation, brainstorming, design thinking, and creative problem-solving. ## About CIS The Creative Intelligence Suite (CIS) extends BMad Method with tools for the fuzzy front-end of development—where ideas are born, problems are reframed, and solutions emerge through structured creativity. ## Modules Included | Agent/Workflow | Purpose | |---------------|---------| | **Innovation Strategist** | Identify disruption opportunities and business model innovation | | **Design Thinking Coach** | Human-centered design through empathy, ideation, and prototyping | | **Brainstorming Coach** | Facilitate creative ideation sessions with proven techniques | | **Problem Solver** | Systematic problem diagnosis and root cause analysis | | **Creative Problem Solver** | Generate creative solutions using lateral thinking | | **Storyteller** | Craft compelling narratives for products and features | | **Presentation Master** | Structure and deliver persuasive presentations | ## Installation CIS is installed as a module during BMad Method setup: ```bash npx bmad-method@alpha install ``` Select **Creative Intelligence Suite** from the modules list. ## Quick Start After installing BMad Method with CIS, try these workflows: ``` /cis-brainstorm # Generate ideas with structured techniques /cis-design-thinking # Human-centered design process /cis-problem-solve # Systematic problem analysis /cis-innovation # Business model and disruption analysis ``` ## When to Use CIS | Situation | Use This | |-----------|----------| | Stuck on a problem | `/cis-problem-solve` | | Need fresh ideas | `/cis-brainstorm` | | Designing for users | `/cis-design-thinking` | | Finding market gaps | `/cis-innovation` | | Telling your product story | `/cis-storytelling` | | Preparing a pitch | `/cis-presentation` | ## Example: Brainstorming Session ``` You: /cis-brainstorm CIS: What would you like to brainstorm about? You: Ways to improve user onboarding CIS: Let's use the SCAMPER te [ASSISTANT] {"what_it_is":"A BMad Method module offering creative thinking agents and workflows for brainstorming, design thinking, and problem-solving via slash commands.","stack":["JavaScript","Python","BMad Method","npm"],"why_mike_cares":"Adjacent to Mike's agent ecosystem and skill/slash-command patterns, but creative ideation isn't a core gap in his SEO/automation stack.","verdict":"worth-knowing","verdict_reason":"BMad Method's module pattern and the SCAMPER-style ideation workflows could inspire skills for Oliver/brainstorming, but it doesn't directly plug into Mike's active agent fleet."}

[USER] updat ehow long this will take [ASSISTANT] Standing by.

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: VoltAgent/awesome-openclaw-skills Stars: 48903 Language: Topics: agent-skills, awesome, awesome-list, awesome-lists, clawd, clawdbot, clawdbot-skill, clawdhub, moltbot, moltbot-skills, openclaw, openclaw-skills Description: The awesome collection of OpenClaw skills. 5,400+ skills filtered and categorized from the official OpenClaw Skills Registry.🦞 README (first 3000 chars): <div align="center"> <a href="https://clawskills.sh/"> <img width="1500" height="500" alt="social" src="https://github.com/user-attachments/assets/a6f310af-8fed-4766-9649-b190575b399d" /> </a> <br/> <br/> <div align="center"> <strong>Discover 5200+ community-built OpenClaw skills, organized by category. </strong> <br /> <br /> </div> [](https://awesome.re) [](#table-of-contents) [](https://github.com/VoltAgent/awesome-clawdbot-skills/pulls?q=is%3Apr+is%3Amerged+sort%3Aupdated-desc) <a href="https://github.com/VoltAgent/voltagent"> <img alt="VoltAgent" src="https://cdn.voltagent.dev/website/logo/logo-2-svg.svg" height="20" /> </a> [](https://s.voltagent.dev/discord) </div> <div align="center"> <strong>More awesome collections for developers</strong> <br /> <br /> </div> <div align="center"> [](https://github.com/VoltAgent/awesome-agent-skills) [](https://github.com/VoltAgent/awesome-claude-code-subagents) [![Codex Subagents][codex-badge]][codex-link] [](https://github.com/VoltAgent/awesome-ai-agent-papers) </div> </div> </div> # Awesome OpenClaw Skills OpenClaw is a locally-running AI assistant that operates directly on your machine. Skills extend its capabilities, allowing it to interact with external services, automate workflows, and perform specialized tasks. This collection helps you discover and install the right skills for your needs. It can also serve as a source of inspiration for OpenClaw use cases. Skills in this list are sourced from ClawHub (OpenClaw's public skills registry) and categorized for easier discovery. ### Installation #### ClawHub CLI ```bash clawhub install <skill-slug> ``` #### Manual Installation Copy the skill folder to one of these locations: | Location | Path | |----------|------| | Global | `~/.opencla [ASSISTANT] { "what_it_is": "A curated awesome-list of 5,200+ community-built OpenClaw skills sourced from ClawHub, categorized for discovery and inspiration.", "stack": ["OpenClaw", "ClawHub CLI", "Markdown", "Agent Skills"], "why_mike_cares": "Mike runs OpenClaw agents on his Mac (mikeybotzmerlino account) and operates a multi-agent ecosystem where skill discovery directly feeds his Fleet and meta-tooling workflows.", "verdict": "check-it-out", "verdict_reason": "Direct overlap with Mike's OpenClaw runtime and skill-curation patterns; a 5K+ skill registry is a high-leverage browse target for his agent ecosystem." }

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: affaan-m/Behavioral_RL Stars: 23 Language: HTML Topics: behavioral-economics, machine-learning, python, reinforcement-learning, research Description: Reinforcement Learning with human behavioral biases integration README (first 3000 chars): # Risk-Sensitive Reinforcement Learning for the Iowa Gambling Task This project implements a risk-sensitive reinforcement learning approach to model human decision-making behavior in the Iowa Gambling Task (IGT). The implementation closely follows the original experimental parameters from Bechara et al. (1994) and incorporates prospect theory and conditional value at risk (CVaR) to model human-like risk sensitivity. ## Experimental Design ### Iowa Gambling Task Parameters Based on Bechara et al. (1994): - Number of trials: 200 (two phases of 100 trials each) - Deck configurations: - Deck A (High risk, high punishment): - Reward: +100 per selection - Punishment: -150 to -350 (frequency: 50%) - Net expected value: -25 per card - Deck B (High risk, infrequent punishment): - Reward: +100 per selection - Punishment: -1250 (frequency: 10%) - Net expected value: -25 per card - Deck C (Low risk, low reward): - Reward: +50 per selection - Punishment: -50 (frequency: 50%) - Net expected value: +25 per card - Deck D (Low risk, infrequent punishment): - Reward: +50 per selection - Punishment: -250 (frequency: 10%) - Net expected value: +25 per card ### Risk-Sensitive Model Parameters 1. Prospect Theory Parameters (based on Tversky & Kahneman, 1992): - α (value function curvature for gains): 0.88 - β (value function curvature for losses): 0.88 - λ (loss aversion coefficient): 2.25 - Reference point: Dynamic, updated based on running average 2. Conditional Value at Risk (CVaR) Parameters: - α (confidence level): 0.05 - λ_risk (risk sensitivity): 0.7 - Window size: 20 trials 3. Learning Parameters: - Learning rate (α): 0.1 - Discount factor (γ): 0.95 - Exploration rate (ε): Linear decay from 1.0 to 0.1 - Batch size: 32 - Memory buffer size: 10000 - Target network update frequency: 100 steps ## Methodology ### 1. Environment Implementation - Custom IGT environment following OpenAI Gym interface - State space: [last_reward, running_average, deck_frequencies] - Action space: Discrete(4) representing decks A-D - Reward structure matching Bechara et al. (1994) ### 2. Model Architecture 1. Baseline Model: - Standard DQN with 3-layer neural network - Layer sizes: [64, 128, 64] - ReLU activation - Adam optimizer (lr=0.001) 2. Risk-Sensitive Model: - Modified DQN incorporating prospect theory value function - CVaR risk measure in Q-value computation - Same architecture as baseline - Additional risk-processing layers ### 3. Training Procedure 1. Phase 1 (Exploration): Episodes 1-100 - Higher exploration rate (ε: 1.0 → 0.3) - Focus on learning deck characteristics - More weight on immediate rewards 2. Phase 2 (Exploitation): Episodes 101-200 - Lower explo [ASSISTANT] {"what_it_is": "Academic reinforcement learning project applying prospect theory and CVaR to model human risk-sensitive decision-making on the Iowa Gambling Task.", "stack": ["Python", "PyTorch", "OpenAI Gym", "DQN"], "why_mike_cares": "Low overlap, this is behavioral economics research, not agent tooling or SEO automation.", "verdict": "skip", "verdict_reason": "Pure academic RL research with no connection to Mike's agent ecosystem, LLM tooling, or marketing automation work."}

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: get-convex/agent Stars: 328 Language: TypeScript Topics: Description: Build AI agents on Convex with persistent chat history README (first 3000 chars): # Convex Agent Component [](https://badge.fury.io/js/@convex-dev%2fagent) Convex provides powerful building blocks for building agentic AI applications, leveraging Components and existing Convex features. With Convex, you can separate your long-running agentic workflows from your UI, without the user losing reactivity and interactivity. ```sh npm i @convex-dev/agent ``` <!-- START: Include on https://convex.dev/components --> AI Agents, built on Convex. [Check out the docs here](https://docs.convex.dev/agents). The Agent component is a core building block for building AI agents. It manages threads and messages, around which you Agents can cooperate in static or dynamic workflows. - [Agents](https://docs.convex.dev/agents/agent-usage) provide an abstraction for using LLMs to represent units of use-case-specific prompting with associated models, prompts, [Tool Calls](https://docs.convex.dev/agents/tools), and behavior in relation to other Agents, functions, APIs, and more. - [Threads](https://docs.convex.dev/agents/threads) persist [messages](https://docs.convex.dev/agents/messages) and can be shared by multiple users and agents (including [human agents](https://docs.convex.dev/agents/human-agents)). - Streaming text and objects using deltas over websockets so all clients stay in sync efficiently, without http streaming. Enables streaming from async functions. - [Conversation context](https://docs.convex.dev/agents/context) is automatically included in each LLM call, including built-in hybrid vector/text search for messages in the thread and opt-in search for messages from other threads (for the same specified user). - [RAG](https://docs.convex.dev/agents/rag) techniques are supported for prompt augmentation from other sources, either up front in the prompt or as tool calls. Integrates with the [RAG Component](https://www.convex.dev/components/rag), or DIY. - [Workflows](https://docs.convex.dev/agents/workflows) allow building multi-step operations that can span agents, users, durably and reliably. - [Files](https://docs.convex.dev/agents/files) are supported in thread history with automatic saving to [file storage](https://docs.convex.dev/file-storage) and ref-counting. - [Debugging](https://docs.convex.dev/agents/debugging) is enabled by callbacks, the [agent playground](https://docs.convex.dev/agents/playground) where you can inspect all metadata and iterate on prompts and context settings, and inspection in the dashboard. - [Usage tracking](https://docs.convex.dev/agents/usage-tracking) is easy to set up, enabling usage attribution per-provider, per-model, per-user, per-agent, for billing & more. - [Rate limiting](https://docs.convex.dev/agents/rate-limiting), powered by the [Rate Limiter Component](https://www.convex.dev/components/rat [ASSISTANT] { "what_it_is": "A Convex component for building AI agents with persistent chat threads, streaming, tool calls, RAG, and durable multi-step workflows.", "stack": ["TypeScript", "Convex", "AI SDK", "WebSockets", "Vector Search"], "why_mike_cares": "Direct overlap: Mike uses Convex in his stack and builds AI agents with persistent memory, threads, tool calls, and workflow orchestration (Hindsight, HawkeyePanel, agent fleet).", "verdict": "check-it-out", "verdict_reason": "First-party Convex agent primitive that maps cleanly onto Mike's existing Convex + agent memory work and could replace custom thread/message plumbing." }

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: mbailey/skills Stars: 1 Language: Python Topics: Description: Claude Code Plugin Marketplace README (first 3000 chars): # Mike Bailey's Skills Marketplace A Claude Code plugin marketplace with skills to somewhat improve things. ## Installation ### Add Claude Code Marketplace ``` claude plugin marketplace add mbailey/skills ``` ### Install a Plugin ``` claude plugin install --scope user show-me@mbailey ``` ### Browse and Install Plugins ``` claude /plugin ``` ## Available Plugins | Plugin | Description | | ------------------------------------------------- | ------------------------------------------------------ | | [show-me](https://github.com/mbailey/show-me) | Let Claude show you files and web pages (tmux, neovim) | | [VoiceMode](https://github.com/mbailey/voicemode) | Natrual conversations with Claude Code | ## Security scanning `scripts/scan_plugin.py` is a deterministic, dependency-free security scanner for plugin directories. It flags unicode tricks (bidi/zero-width/homoglyphs), network access, destructive commands, pipe-to-shell, credential-path access, encoded payloads, privilege escalation, and compiled bytecode. Run it against a plugin directory (or a `plugins/` parent) with [`uv`](https://docs.astral.sh/uv/): ``` uv run scripts/scan_plugin.py <plugin-dir> uv run scripts/scan_plugin.py plugins/ # scan every plugin uv run scripts/scan_plugin.py <dir> --format=markdown ``` Exit codes: `0` clean, `1` usage error, `2` BLOCK findings, `3` WARN only. CI runs this as a **non-blocking, informational** job (it never gates merges). The marketplace currently lists external `url`-source plugins with no bundled plugin directories, so the job self-skips until skills are vendored under `plugins/`. The scanner is vendored from [trailofbits/skills-curated](https://github.com/trailofbits/skills-curated) (`scripts/scan_plugin.py`) and is licensed under [CC-BY-SA-4.0](https://creativecommons.org/licenses/by-sa/4.0/); see the attribution header in the file. Analyze this repo per your instructions. Return ONLY valid JSON, no markdown fences. [ASSISTANT] {"what_it_is": "A Claude Code plugin marketplace by Mike Bailey hosting skills like show-me and VoiceMode, with a bundled security scanner for plugin directories.", "stack": ["Claude Code", "Python", "uv"], "why_mike_cares": "Mike runs a heavy Claude Code skills/plugins ecosystem and the bundled scan_plugin.py from trailofbits is directly useful for vetting third-party skills before installing them globally.", "verdict": "worth-knowing", "verdict_reason": "The marketplace itself is thin (2 plugins, 1 star), but the vendored security scanner is a drop-in tool for Mike's CLI-first routing rule about reviewing generated SKILL.md and install scripts before adding to global skills."}

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: CopilotKit/fullstack-solution-template-for-agentcore Stars: 1 Language: Python Topics: Description: Flexible Fullstack solution template for production-ready deployments of any use case on Amazon Bedrock AgentCore. README (first 3000 chars): # Fullstack AgentCore Solution Template (FAST) _Author's note: for the official name for this solution is the "Fullstack Solution Template for Agentcore" but it is referred to throughout this code base as FAST for convenience._ The Fullstack AgentCore Solution Template (FAST) is a starter project repository that enables users (delivery scientists and engineers) to quickly deploy a secured, web-accessible React frontend connected to an AgentCore backend. Its purpose is to accelerate building full stack applications on AgentCore from weeks to days by handling the undifferentiated heavy lifting of infrastructure setup and to enable vibe-coding style development on top. The only central dependency of FAST is AgentCore. It is agnostic to agent SDK (Strands, LangGraph, etc) and to coding assistant platforms (Q, Kiro, Cline, Claude Code, etc). FAST is designed with security and vibe-codability as primary tenets. Best practices and knowledge from experts are codified in _documentation_ in this repository rather than in _code_. By including this documentation in an AI coding assistant's context, or by instructing the AI coding assistant to leverage best practices and code snippets found in the documentation, delivery scientists and developers can quickly vibe-build AgentCore applications for any use case. AI coding assistants can be used to fully customize the frontend and the infrastructure, enabling scientists to focus the areas where their knowledge is most impactful: the actual prompt engineering and GenAI implementation details. With FAST as a starting point and development framework, delivery scientists and engineers will accelerate their development process and deliver production quality AgentCore code following architecture and security best practices without having to learn any frontend or infrastructure code. ## FAST Baseline System FAST comes deployable out-of-the-box with a fully functioning, full-stack application. This application represents starts as a basic multi-turn chat agent where the backend agent has access to tools. **Do not let this deter you, even if your use case is entirely different! If your application requires AgentCore, customizing FAST to any use case is extremely straightforward. That is the intended use of FAST!** The application is intentionally kept very, very simple to allow developers to easily build up whatever they want on top of the baseline. The tools shipped out of the box include: 1. **Gateway Tools** - Lambda-based tools behind AgentCore Gateway with authentication: - Text analysis tool (counts words and letter frequency) 2. **Code Interpreter** - Direct integration with Amazon Bedrock AgentCore Code Interpreter: - Secure Python code execution in isolated sandbox - Session management with state persistence - [ASSISTANT] {"what_it_is": "Fullstack starter template that deploys a React frontend wired to an Amazon Bedrock AgentCore backend with auth, infra, and tool integration scaffolded out.", "stack": ["Python", "React", "AWS Bedrock AgentCore", "AWS Lambda", "AWS CDK"], "why_mike_cares": "Low overlap, Mike runs Claude/Anthropic-native agent stacks on Vercel/Supabase, not AWS Bedrock AgentCore.", "verdict": "skip", "verdict_reason": "AWS Bedrock-locked template with zero traction (1 star) and no fit with Mike's Claude Code, Next.js, or MCP-based agent ecosystem."}

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>19G total. Waiting.</result> <usage><total_tokens>91178</total_tokens><tool_uses>0</tool_uses><duration_ms>2923</duration_ms></usage> </task-notification> [ASSISTANT] Standing by.

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: GoogleCloudPlatform/asl-ml-immersion Stars: 2548 Language: Jupyter Notebook Topics: adk, adk-pyth, agents, generative-ai, google-cloud-platform, machine-learning, tensorflow Description: Notebooks, code samples and reference for machine learning and generative ai on Google Cloud for the Advanced Solutions Lab (ASL) bootcamps. README (first 3000 chars): # Advanced Solutions Lab ## Overview This repository contains AI and Machine Learning contents meant to be run on Google Cloud. This is maintained by Google Cloud’s [Advanced Solutions Lab (ASL)](https://cloud.google.com/asl) team. This repository contains 3 main modules to covers various AI/ML toipcs: - `asl_core`: A wide range of model architectures (DNN, CNN, RNN, transformers, SNGP, etc.) targeting many data modalities (tabular, image, text, time-series) implemented mainly in Tensorflow and Keras. - `asl_mlops`: Tools on Google Cloud’s Vertex AI for operationalizing Tensorflow, Scikit-learn and PyTorch models at scale (e.g. Vertex training, tuning, and serving, TFX and Kubeflow pipelines). - `asl_genai`: Generative AI and Agent System using Gemini and Agentic Frameworks like Google ADK. ## Repository Structure Each module (`asl_core`, `asl_mlops`, `asl_genai`) has separate environment and materials, which are organized in each directory. All learning materials are in the contets folder. This folder is organized by different topics. Each folder contains a `labs` and a `solutions` folder. Use the `labs` notebooks to test your coding skills by filling in TODOs and refer to the notebooks in the `solutions` folder to verify your code. We have three main folders described below: ``` ├── asl_core │ ├── notebooks - contains learning materials organized by topic │ │ ├── building_production_ml_systems │ │ │ ├── labs │ │ │ └── solutions │ │ ├── end-to-end-structured │ │ ├── image_models │ │ ├── ... │ ├── kernels - contains kernel scripts needed for certain notebooks │ ├── scaffolds - contains sample code to accelerate AI/ML projects │ ├── requirements.txt - dependencies for this module ├── asl_mlops │ ├── ... ├── asl_genai │ ├── ... ├── ... ``` ## Environment Setup ### Step 1. Run the Setup Script on Cloud Shell This repository is tested on Vertex AI Workbench and Cloud Workstations. To begin, run the setup script in [Cloud Shell](https://shell.cloud.google.com) to configure essential project infrastructure (APIs, IAM, Buckets). Run the setup script in [Cloud Shell](https://shell.cloud.google.com) to provision your environment. ```bash git clone https://github.com/GoogleCloudPlatform/asl-ml-immersion.git cd asl-ml-immersion bash scripts/setup_env.sh ``` You will be prompted to select the environment to set up: * **1) Vertex AI Workbench:** Setup Vertex AI Workbench. * **2) Cloud Workstations:** Setup Cloud Workstations. * **3) Setup both:** Setup both environments. * **4) Skip:** Setup project infrastructure (APIs, IAM, Buckets) only. By selecting the option 1-3, you can automatically setup the environment, [ASSISTANT] {"what_it_is":"Google Cloud's Advanced Solutions Lab bootcamp repo with notebooks covering TensorFlow ML, Vertex AI MLOps, and Gemini/ADK generative AI labs.","stack":["Jupyter","TensorFlow","Keras","Vertex AI","Gemini","Google ADK","Python"],"why_mike_cares":"The asl_genai module covers Google ADK agent framework which is adjacent to Mike's multi-agent ecosystem work, but the GCP-heavy stack doesn't match his Next.js/Supabase/Claude-centric setup.","verdict":"worth-knowing","verdict_reason":"ADK agent patterns and Gemini agentic examples are useful reference for Mike's agent fleet, but bootcamp notebooks tied to Vertex AI infra have low direct reuse."}

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: privatenumber/cleye Stars: 626 Language: TypeScript Topics: argv, cli, commandline-interface, commandline-parser, flags, node, parameters, typed, typescript Description: 👁‍🗨 Strongly typed CLI development for Node.js README (first 3000 chars): <p align="center"> <img width="110" src=".github/logo.webp"> </p> <h1 align="center"> cleye <br> <a href="https://npm.im/cleye"><img src="https://badgen.net/npm/v/cleye"></a> <a href="https://npm.im/cleye"><img src="https://badgen.net/npm/dm/cleye"></a> </h1> The intuitive command-line interface (CLI) development tool. ### Features - Minimal API surface - Powerful flag parsing - Strongly typed parameters and flags - Command support - Help documentation generation (customizable too!) → [Try it out online](https://stackblitz.com/edit/cleye-demo?devtoolsheight=50&file=examples/greet.ts&view=editor) <br> <p align="center"> <a href="https://github.com/sponsors/privatenumber/sponsorships?tier_id=398771"><img width="412" src="https://raw.githubusercontent.com/privatenumber/sponsors/master/banners/assets/donate.webp"></a> <a href="https://github.com/sponsors/privatenumber/sponsorships?tier_id=397608"><img width="412" src="https://raw.githubusercontent.com/privatenumber/sponsors/master/banners/assets/sponsor.webp"></a> </p> <p align="center">^{<i>Already a sponsor?</i> Join the discussion in the <a href="https://github.com/pvtnbr/cleye">Development repo</a>!}</p> ## Install ```bash npm i cleye ``` ## About _Cleye_ makes it very easy to develop command-line scripts in Node.js. It handles argv parsing to give you strongly typed parameters + flags and generates `--help` documentation based on the provided information. Here's an example script that simply logs: `Good morning/evening <name>!`: _greet.js:_ ```ts import { cli } from 'cleye' // Parse argv const argv = cli({ name: 'greet.js', // Define parameters parameters: [ '<first name>', // First name is required '[last name]' // Last name is optional ], // Define flags/options flags: { // Parses `--time` as a string time: { type: String, description: 'Time of day to greet (morning or evening)', default: 'morning' } } }) const name = [argv._.firstName, argv._.lastName].filter(Boolean).join(' ') if (argv.flags.time === 'morning') { console.log(`Good morning ${name}!`) } else { console.log(`Good evening ${name}!`) } ``` 🛠 In development, type hints are provided on parsed flags and parameters: <p align="center"> <br> <img src=".github/typed-flags.png" width="600"> <br> <i>Type hints for Cleye's output are very verbose and readable</i> <br> <br> </p> 📖 Generated help documentation can be viewed with the `--help` flag: ```sh $ node greet.js --help greet.js Usage: greet.js [flags...] <first name> [last name] Flags: -h, --help Show help --time <string> Time of day to greet (morning or evening) (default: "morning") ``` ✅ Run the scrip [ASSISTANT] {"what_it_is":"A TypeScript-first CLI argument parser for Node.js with strongly typed flags, parameters, commands, and auto-generated help docs.","stack":["TypeScript","Node.js"],"why_mike_cares":"Mike builds Python and TypeScript CLI tooling (GSD 2, PrintingPress, CLI-Anything harnesses, brv) where typed flags and auto-help would tighten the agent-native CLI build standard.","verdict":"worth-knowing","verdict_reason":"Solid pick if any new TS CLI gets built in the ecosystem, but Python dominates current CLI work so it is adjacent, not urgent."}

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: openai/symphony Stars: 24064 Language: Elixir Topics: Description: Symphony turns project work into isolated, autonomous implementation runs, allowing teams to manage work instead of supervising coding agents. README (first 3000 chars): # Symphony Symphony turns project work into isolated, autonomous implementation runs, allowing teams to manage work instead of supervising coding agents. [](.github/media/symphony-demo.mp4) _In this [demo video](.github/media/symphony-demo.mp4), Symphony monitors a Linear board for work and spawns agents to handle the tasks. The agents complete the tasks and provide proof of work: CI status, PR review feedback, complexity analysis, and walkthrough videos. When accepted, the agents land the PR safely. Engineers do not need to supervise Codex; they can manage the work at a higher level._ > [!WARNING] > Symphony is a low-key engineering preview for testing in trusted environments. ## Running Symphony ### Requirements Symphony works best in codebases that have adopted [harness engineering](https://openai.com/index/harness-engineering/). Symphony is the next step -- moving from managing coding agents to managing work that needs to get done. ### Option 1. Make your own Tell your favorite coding agent to build Symphony in a programming language of your choice: > Implement Symphony according to the following spec: > https://github.com/openai/symphony/blob/main/SPEC.md ### Option 2. Use our experimental reference implementation Check out [elixir/README.md](elixir/README.md) for instructions on how to set up your environment and run the Elixir-based Symphony implementation. You can also ask your favorite coding agent to help with the setup: > Set up Symphony for my repository based on > https://github.com/openai/symphony/blob/main/elixir/README.md --- ## License This project is licensed under the [Apache License 2.0](LICENSE). Analyze this repo per your instructions. Return ONLY valid JSON, no markdown fences. [ASSISTANT] {"what_it_is": "OpenAI's autonomous coding agent orchestrator that spawns isolated implementation runs from Linear tickets, with built-in proof-of-work (CI, PR reviews, walkthrough videos) so engineers manage work instead of babysitting Codex.", "stack": ["Elixir", "Codex", "Linear API", "GitHub PRs", "CI/CD"], "why_mike_cares": "Direct overlap with Mike's Carlos/Oliver orchestration model, ClawControl task dispatch, and his 'manage work not agents' philosophy. This is OpenAI's official take on the exact pattern he's building.", "verdict": "check-it-out", "verdict_reason": "OpenAI's reference architecture for autonomous agent orchestration with proof-of-work, mirroring Mike's Oliver/Carlos/POWD pattern. Worth studying the SPEC.md even if he ignores the Elixir impl."}

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: NVIDIA-AI-Blueprints/biomedical-aiq-research-agent Stars: 128 Language: Python Topics: Description: README (first 3000 chars): # NVIDIA Biomedical AI-Q Research Agent Developer Blueprint ## Overview The NVIDIA [Biomedical AI-Q Research Agent Developer Blueprint](https://build.nvidia.com/nvidia/biomedical-aiq-research-agent) allows you to create a deep research agent with virtual screening capabilities that can run on-premise, allowing anyone to create detailed research reports using on-premise data and web search. This developer blueprint is built on top of the [AI-Q NVIDIA Research Assistant Blueprint](https://build.nvidia.com/nvidia/aiq). We have added capabilities from the [Virtual Screening Blueprint](https://build.nvidia.com/nvidia/generative-virtual-screening-for-drug-discovery) in addition to the research capability, so that when a biomedical researcher is investigating a condition or disease, with a target protein and recent small-molecule therapy, virtual screening can help with discovering novel small-molecule therapies through guided molecular generation and docking. This developer blueprint serves as a demonstration on how you could add in your own custom functionalities that may not be virtual screening, to the existing research and report generation functionalities in the foundational [AI-Q NVIDIA Research Assistant Blueprint](https://build.nvidia.com/nvidia/aiq). ## Table of Contents - [Key Features](#key-features) - [Target Audience](#target-audience) - [Software Components](#software-components) - [Technical Diagram](#technical-diagram) - [Minimum System Requirements](#minimum-system-requirements) - [OS Requirements](#os-requirements) - [Deploy Options](#deploy-options) - [Drivers](#drivers) - [Hardware Requirements](#hardware-requirements) - [API Keys](#api-keys) - [Getting Started](#getting-started) - [License](#license) - [Security Considerations](#security-considerations) ## Key Features - **Deep Research**: Given a report topic and desired report structure, an agent (1) creates a report plan, (2) searches data sources for answers, (3) writes a report, (4) reflects on gaps in the report for further queries, (5) finishes a report with a list of sources. - **Parallel Search**: During the research phase, multiple research questions are searched in parallel. For each query, the RAG service is consulted and an LLM-as-a-judge is used to check the relevancy of the results. If more information is needed, a fallback web search is performed. This search approach ensures internal documents are given preference over generic web results while maintaining accuracy. Performing query search in parallel allows for many data sources to be consulted in an efficient manner. - **Human-in-the-loop**: Human feedback on the report plan, interactive report edits, and Q&A with the final report. - **Data Sources**: Integration with the NVIDIA RAG blueprint to search multimodal documents with text, charts, and tables. For a full list of file formats supported such as `pdf`, `pptx`, `docx`, `jpeg [ASSISTANT] { "what_it_is": "NVIDIA's biomedical deep-research agent blueprint that combines parallel RAG search, human-in-the-loop report editing, and drug-discovery virtual screening on-prem.", "stack": ["Python", "NVIDIA AI-Q", "NVIDIA RAG Blueprint", "LLM-as-judge", "On-prem GPU"], "why_mike_cares": "The parallel-search-plus-judge-plus-web-fallback pattern and human-in-the-loop report editing mirror Mike's SEO research agents (Einstein, Sherlock, Raven) and Merlino Magic Blog pipeline.", "verdict": "worth-knowing", "verdict_reason": "Biomedical/drug-discovery focus is off-domain, but the deep-research orchestration pattern (plan, parallel query, judge, reflect, finalize) is directly steal-able for Mike's SEO and content research agents." }

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>Waiting.</result> <usage><total_tokens>91048</total_tokens><tool_uses>0</tool_uses><duration_ms>3169</duration_ms></usage> </task-notification> [ASSISTANT] Standing by.

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: NVIDIA-AI-Blueprints/Multi-Agent-Intelligent-Warehouse Stars: 84 Language: Python Topics: Description: README (first 3000 chars): # Multi-Agent-Intelligent-Warehouse *NVIDIA Blueprint–aligned multi-agent assistant for warehouse operations.* [](https://opensource.org/licenses/Apache-2.0) [](https://www.python.org/downloads/) [](https://fastapi.tiangolo.com/) [](https://reactjs.org/) [](https://www.nvidia.com/en-us/ai-data-science/nim/) [](https://www.postgresql.org/) [](https://milvus.io/) [](https://www.docker.com/) [](https://prometheus.io/) [](https://grafana.com/) ## Table of Contents - [Overview](#overview) - [Acronyms & Abbreviations](#acronyms--abbreviations) - [System Architecture](#system-architecture) - [Key Features](#key-features) - [Quick Start](#quick-start) - [Multi-Agent System](#multi-agent-system) - [API Reference](#api-reference) - [Monitoring & Observability](#monitoring--observability) - [NeMo Guardrails](#nemo-guardrails) - [Development Guide](#development-guide) - [Contributing](#contributing) - [License](#license) ## Acronyms & Abbreviations | Acronym | Definition | |---------|------------| | **ADR** | Architecture Decision Record | | **API** | Application Programming Interface | | **BOL** | Bill of Lading | | **cuML** | CUDA Machine Learning | | **cuVS** | CUDA Vector Search | | **EAO** | Equipment & Asset Operations (Agent) | | **ERP** | Enterprise Resource Planning | | **GPU** | Graphics Processing Unit | | **HTTP/HTTPS** | Hypertext Transfer Protocol (Secure) | | **IoT** | Internet of Things | | **JSON** | JavaScript Object Notation | | **JWT** | JSON Web Token | | **KPI** | Key Performance Indicator | | **LLM** | Large Language Model | | **LOTO** | Lockout/Tagout | | **MAPE** | Mean Absolute Percentage Error | | **MCP** | Model Context Protocol | | **NeMo** | NVIDIA NeMo | | **NIM/NIMs** | NVIDIA Inference Microservices | | **OCR** | Optical Character Recognition | | **PPE** | Personal Protective Equipment | | **QPS** | Queries Per Second | | **RAG** | Retrieval-Augmented Generation | | **RAPIDS** | Rapid Analytics Platform for Interactive Data Science | | **RBAC** | Role-Based Access Control | | **RFID** | Radio Frequency Identification | | **RMSE** | Root Mean Square Error | | **REST** | Representational State Transfer | | **SDS** | Safety Data Sheet | | **SKU** | Stock Keeping Uni [ASSISTANT] {"what_it_is": "NVIDIA blueprint for a multi-agent AI assistant tailored to warehouse operations, built on NIMs, FastAPI, React, Milvus, and NeMo Guardrails.", "stack": ["Python", "FastAPI", "React", "NVIDIA NIMs", "NeMo Guardrails", "PostgreSQL", "Milvus", "Docker", "Prometheus", "Grafana", "MCP"], "why_mike_cares": "Reference architecture for production multi-agent systems with MCP, guardrails, and observability, patterns transferable to Mike's agent fleet (Oliver/Carlos/leads) even though warehouse domain is irrelevant.", "verdict": "worth-knowing", "verdict_reason": "Solid multi-agent + MCP + guardrails reference architecture from NVIDIA, but warehouse-specific and GPU/NIM-heavy, so it's a pattern source not a drop-in tool for Mike's stack."}

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: obra/superpowers Stars: 196041 Language: Shell Topics: Description: An agentic skills framework & software development methodology that works. README (first 3000 chars): # Superpowers Superpowers is a complete software development methodology for your coding agents, built on top of a set of composable skills and some initial instructions that make sure your agent uses them. ## Quickstart Give your agent Superpowers: [Claude Code](#claude-code), [Codex CLI](#codex-cli), [Codex App](#codex-app), [Factory Droid](#factory-droid), [Gemini CLI](#gemini-cli), [OpenCode](#opencode), [Cursor](#cursor), [GitHub Copilot CLI](#github-copilot-cli). ## How it works It starts from the moment you fire up your coding agent. As soon as it sees that you're building something, it *doesn't* just jump into trying to write code. Instead, it steps back and asks you what you're really trying to do. Once it's teased a spec out of the conversation, it shows it to you in chunks short enough to actually read and digest. After you've signed off on the design, your agent puts together an implementation plan that's clear enough for an enthusiastic junior engineer with poor taste, no judgement, no project context, and an aversion to testing to follow. It emphasizes true red/green TDD, YAGNI (You Aren't Gonna Need It), and DRY. Next up, once you say "go", it launches a *subagent-driven-development* process, having agents work through each engineering task, inspecting and reviewing their work, and continuing forward. It's not uncommon for Claude to be able to work autonomously for a couple hours at a time without deviating from the plan you put together. There's a bunch more to it, but that's the core of the system. And because the skills trigger automatically, you don't need to do anything special. Your coding agent just has Superpowers. ## Sponsorship If Superpowers has helped you do stuff that makes money and you are so inclined, I'd greatly appreciate it if you'd consider [sponsoring my opensource work](https://github.com/sponsors/obra). Thanks! - Jesse ## Installation Installation differs by harness. If you use more than one, install Superpowers separately for each one. ### Claude Code Superpowers is available via the [official Claude plugin marketplace](https://claude.com/plugins/superpowers) #### Official Marketplace - Install the plugin from Anthropic's official marketplace: ```bash /plugin install superpowers@claude-plugins-official ``` #### Superpowers Marketplace The Superpowers marketplace provides Superpowers and some other related plugins for Claude Code. - Register the marketplace: ```bash /plugin marketplace add obra/superpowers-marketplace ``` - Install the plugin from this marketplace: ```bash /plugin install superpowers@superpowers-marketplace ``` ### Codex CLI Superpowers is available via the [official Codex plugin marketplace](https://github.com/openai/plugins). - Open the plugin search interface: ```bash /plugins ``` - Search for Superpowers: ```bash su [ASSISTANT] {"what_it_is": "An agentic skills framework and TDD-driven software development methodology that adds spec-first planning, subagent-driven execution, and composable skills to coding agents.","stack": ["Claude Code","Codex","Shell","Skills Framework","Subagent Orchestration","TDD"],"why_mike_cares": "Directly overlaps with Mike's Superpowers-based Claude Code stack, his Oliver/Carlos orchestrator pattern, subagent dispatch protocol, and skill-driven workflows already in production.","verdict": "check-it-out","verdict_reason": "This is the upstream of the exact skills system Mike already runs daily across Claude Code, OpenClaw, and Codex lanes."}

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: obra/lace Stars: 22 Language: TypeScript Topics: Description: Lightweight agentic coding environment README (first 3000 chars): Analyze this repo per your instructions. Return ONLY valid JSON, no markdown fences. [ASSISTANT] {"what_it_is":"Lightweight agentic coding environment built in TypeScript, positioned as a minimal alternative to heavier coding agent frameworks.","stack":["TypeScript","Node.js"],"why_mike_cares":"Direct overlap with Mike's agentic coding stack (Claude Code, Codex, OpenClaw) and his interest in minimalist, zero-dependency agent harnesses.","verdict":"worth-knowing","verdict_reason":"Low star count and no README detail provided, but the agentic coding niche matches Mike's daily tooling, worth a peek without urgent action."}

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: disler/claude-code-hooks-mastery Stars: 3682 Language: Python Topics: Description: Master Claude Code Hooks README (first 3000 chars): # Claude Code Hooks Mastery [Claude Code Hooks](https://docs.anthropic.com/en/docs/claude-code/hooks) - Quickly master how to use Claude Code hooks to add deterministic (or non-deterministic) control over Claude Code's behavior. Plus learn about [Claude Code Sub-Agents](#claude-code-sub-agents), the powerful [Meta-Agent](#the-meta-agent), and [Team-Based Validation](#team-based-validation-system) with agent orchestration. <img src="images/hooked.png" alt="Claude Code Hooks" style="max-width: 800px; width: 100%;" /> ## Table of Contents - [Prerequisites](#prerequisites) - [Hook Lifecycle & Payloads](#hook-lifecycle--payloads) - [What This Shows](#what-this-shows) - [UV Single-File Scripts Architecture](#uv-single-file-scripts-architecture) - [Key Files](#key-files) - [Features Demonstrated](#features-demonstrated) - [Hook Error Codes & Flow Control](#hook-error-codes--flow-control) - [UserPromptSubmit Hook Deep Dive](#userpromptsubmit-hook-deep-dive) - [Claude Code Sub-Agents](#claude-code-sub-agents) - [Team-Based Validation System](#team-based-validation-system) - [Output Styles Collection](#output-styles-collection) - [Custom Status Lines](#custom-status-lines) ## Prerequisites This requires: - **[Astral UV](https://docs.astral.sh/uv/getting-started/installation/)** - Fast Python package installer and resolver - **[Claude Code](https://docs.anthropic.com/en/docs/claude-code)** - Anthropic's CLI for Claude AI ### Optional Setup: Optional: - **[ElevenLabs](https://elevenlabs.io/)** - Text-to-speech provider (with MCP server integration) - **[ElevenLabs MCP Server](https://github.com/elevenlabs/elevenlabs-mcp)** - MCP server for ElevenLabs - **[Firecrawl MCP Server](https://www.firecrawl.dev/mcp)** - Web scraping and crawling MCP server (my favorite scraper) - **[OpenAI](https://openai.com/)** - Language model provider + Text-to-speech provider - **[Anthropic](https://www.anthropic.com/)** - Language model provider - **[Ollama](https://ollama.com/)** - Local language model provider ## Hook Lifecycle & Payloads This demo captures all 13 Claude Code hook lifecycle events with their JSON payloads: ### Hook Lifecycle Overview ```mermaid flowchart TB subgraph SESSION["🟢 Session Lifecycle"] direction TB SETUP[["🔧 Setup<br/>(init/maintenance)"]] START[["▶️ SessionStart<br/>(startup/resume/clear)"]] END[["⏹️ SessionEnd<br/>(exit/sigint/error)"]] end subgraph MAIN["🔄 Main Conversation Loop"] direction TB PROMPT[["📝 UserPromptSubmit"]] CLAUDE["Claude Processes"] subgraph TOOLS["🛠️ Tool Execution"] direction TB PRE[["🔒 PreToolUse"]] PERM[["❓ PermissionRequest"]] EXEC["Tool Executes"] POST[["✅ PostToolUse"]] FAIL[["❌ PostToolUseFailure"]] end subgraph SUBAGENT["🤖 Subagent Lifecycle"] direction TB [ASSISTANT] { "what_it_is": "Comprehensive demo and reference implementation for mastering all 13 Claude Code hook lifecycle events, sub-agents, meta-agents, and team-based validation patterns.", "stack": ["Claude Code", "Python", "UV", "Hooks", "Sub-Agents", "MCP", "ElevenLabs", "Firecrawl"], "why_mike_cares": "Directly overlaps with Mike's Claude Code orchestration layer, agent fleet (Oliver/Carlos/Queen), POWD hooks, auto-validation hooks, and the witness-curator attestation protocol he enforces globally.", "verdict": "check-it-out", "verdict_reason": "Disler is a top-tier Claude Code educator and this is the canonical hooks mastery repo, perfect reference for hardening Mike's existing hook-driven validation and SPARC gate enforcement." }

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>17G total. Waiting.</result> <usage><total_tokens>90897</total_tokens><tool_uses>0</tool_uses><duration_ms>2329</duration_ms></usage> </task-notification> [ASSISTANT] Standing by.

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: disler/pi-vs-claude-code Stars: 1037 Language: TypeScript Topics: Description: Comparison between open source PI agent and closed source Claude Code agent README (first 3000 chars): # pi-vs-cc A collection of [Pi Coding Agent](https://github.com/mariozechner/pi-coding-agent) customized instances. _Why?_ To showcase what it looks like to hedge against the leader in the agentic coding market, Claude Code. Here we showcase how you can customize the UI, agent orchestration tools, safety auditing, agent to agent orchestration, and cross-agent integrations. > Want to see these **6+ unique Pi Agent Harnesses in action?** Watch [Pi Coding Agent: The Only Claude Code Competitor](https://youtu.be/f8cfH5XX-XU). > 🆕 **Pi-to-Pi agent-to-agent communication**. Jump to [Pi-to-Pi Communication](#pi-to-pi-agent-to-agent-communication) or watch [Pi to Pi: Two-Way Agent Orchestration](https://youtu.be/PIdETjcXNIk). <div align="center"> <img src="./images/pi-logo.png" alt="pi-vs-cc" width="700"> </div> --- ## Prerequisites All three are required: | Tool | Purpose | Install | | --------------- | ------------------------- | ---------------------------------------------------------- | | **Bun** ≥ 1.3.2 | Runtime & package manager | [bun.sh](https://bun.sh) | | **just** | Task runner | `brew install just` | | **pi** | Pi Coding Agent CLI | [Pi docs](https://github.com/mariozechner/pi-coding-agent) | --- ## API Keys Pi does **not** auto-load `.env` files — API keys must be present in your shell's environment **before** you launch Pi. A sample file is provided: ```bash cp .env.sample .env # copy the template # open .env and fill in your keys ``` `.env.sample` covers the four most popular providers: | Provider | Variable | Get your key | | ---------------- | -------------------- | ---------------------------------------------------------------------------------------------------------- | | OpenAI | `OPENAI_API_KEY` | [platform.openai.com](https://platform.openai.com/api-keys) | | Anthropic | `ANTHROPIC_API_KEY` | [console.anthropic.com](https://console.anthropic.com/settings/keys) | | Google | `GEMINI_API_KEY` | [aistudio.google.com](https://aistudio.google.com/app/apikey) | | OpenRouter | `OPENROUTER_API_KEY` | [openrouter.ai](https://openrouter.ai/keys) | | Many Many Others | `***` | [Pi Providers docs](https://github.com/badlogic/pi-mono/blob/main/packages/coding-agent/docs/providers.md) | ### Sourcing your keys Pick whichever approach fits your workflow: **Option [ASSISTANT] {"what_it_is":"A collection of customized Pi Coding Agent harnesses showcasing UI customization, agent orchestration, safety auditing, and Pi-to-Pi agent-to-agent communication as an open-source alternative to Claude Code.","stack":["TypeScript","Bun","Pi Coding Agent","just"],"why_mike_cares":"Direct overlap with Mike's agentic coding stack (Claude Code, Oliver/Carlos orchestration, agent-to-agent messaging via claude-peers) and his pattern of building custom agent harnesses.","verdict":"check-it-out","verdict_reason":"Disler is a top-tier agentic coding builder and Pi's customizable harness + agent-to-agent orchestration directly parallels Mike's Oliver/Carlos/leads architecture, worth studying for hedge patterns against Claude Code lock-in."}

[USER] You are an expert technical analyst for Mike Merlino, an AI agency operator and builder. Your job is to evaluate GitHub repositories and extract structured signal. Mike's domains: AI agents, LLM tooling, Claude Code, MCP servers, prompt engineering, SEO automation, GoHighLevel, Next.js, ShadCN, Vercel, Supabase, voice AI, agentic coding, Discord/Telegram bots, scheduler/cron patterns, Python automation, TypeScript. Return ONLY valid JSON: { "what_it_is": "one sentence plain English description", "stack": ["Technology1", "Technology2"], "why_mike_cares": "one sentence on overlap with Mike's work, or 'Low overlap'", "verdict": "check-it-out" | "worth-knowing" | "skip", "verdict_reason": "one sentence" } Verdict: - "check-it-out": direct overlap with Mike's active projects or tools he uses - "worth-knowing": interesting adjacent tool, may matter later - "skip": no clear overlap Be OPINIONATED. Not neutral. Repo: steveyegge/mcp_agent_mail Stars: 47 Language: Python Topics: Description: Like gmail for your coding agents. Lets various different agents communicate and coordinate with each other. README (first 3000 chars): # MCP Agent Mail  > "It's like gmail for your coding agents!" A mail-like coordination layer for coding agents, exposed as an HTTP-only FastMCP server. It gives agents memorable identities, an inbox/outbox, searchable message history, and voluntary file reservation "leases" to avoid stepping on each other. Think of it as asynchronous email + directory + change-intent signaling for your agents, backed by Git (for human-auditable artifacts) and SQLite (for indexing and queries). Status: Under active development. The design is captured in detail in `project_idea_and_guide.md` (start with the original prompt at the top of that file). ## Why this exists Modern projects often run multiple coding agents at once (backend, frontend, scripts, infra). Without a shared coordination fabric, agents: - Overwrite each other's edits or panic on unexpected diffs - Miss critical context from parallel workstreams - Require humans to "liaison" messages across tools and teams This project provides a lightweight, interoperable layer so agents can: - Register a temporary-but-persistent identity (e.g., GreenCastle) - Send/receive GitHub-Flavored Markdown messages with images - Search, summarize, and thread conversations - Declare advisory file reservations (leases) on files/globs to signal intent - Inspect a directory of active agents, programs/models, and activity It's designed for: FastMCP clients and CLI tools (Claude Code, Codex, Gemini CLI, etc.) coordinating across one or more codebases. ## From Idea Spark to Shipping Swarm If a blank repo feels daunting, follow the field-tested workflow we documented in `project_idea_and_guide.md` (“Appendix: From Blank Repo to Coordinated Swarm”): - **Ideate fast:** Write a scrappy email-style blurb about the problem, desired UX, and any must-have stack picks (≈15 minutes). - **Promote it to a plan:** Feed that blurb to GPT-5 Pro (and optionally Grok4 Heavy / Opus 4.1) until you get a granular Markdown plan, then iterate on the plan file while it’s still cheap to change. The Markdown Web Browser sample plan shows the level of detail to aim for. - **Codify the rules:** Clone a tuned `AGENTS.md`, add any tech-specific best-practice guides, and let Codex scaffold the repo plus Beads tasks straight from the plan. - **Spin up the swarm:** Launch multiple Codex panes (or any agent mix), register each identity with Agent Mail, and have them acknowledge `AGENTS.md`, the plan document, and the Beads backlog before touching code. - **Keep everyone fed:** Reuse the canned instruction cadence from the tweet thread or, better yet, let the commercial Companion app’s Message Stacks broadcast those prompts automatically so you never hand-feed panes again. Watch the full 23-minute walkthrough (https://youtu.be/68VVcqMED [ASSISTANT] {"what_it_is": "An HTTP-based FastMCP server that gives coding agents email-like coordination — identities, inboxes, searchable message history, and file reservation leases to prevent edit conflicts.", "stack": ["Python", "FastMCP", "SQLite", "Git", "MCP"], "why_mike_cares": "Directly overlaps Mike's multi-agent orchestration stack (Oliver/Carlos/leads, claude-peers, OpenClaw, parallel worktree agents) — solves the exact coordination problem his fleet hits when multiple agents touch the same codebase.", "verdict": "check-it-out", "verdict_reason": "Steve Yegge building an MCP coordination layer for parallel coding agents is a near-perfect fit for Mike's Paperclip/Fleet/Conductor pattern and worth stealing patterns from immediately."}

[USER] is this helpful at all for me if its just code? [ASSISTANT] **Honest answer: no, not really.** Here's why it's a dud for you specifically: - **Stack mismatch**: it's plumbing for Vite SPAs. Next.js gives you all of it natively (`generateMetadata`, route handlers for sitemap.xml, JSON-LD in layout). You'd be installing crutches you don't need. - **You already have it**: `technical-seo-agent`, `schema-generator`, `schema-spider`, `auto-schema-from-template`, `core-web-vitals`, `einstein` agent. They cover meta tags, OG, canonical, sitemap, robots, JSON-LD, CWV. - **Doesn't write content**: zero overlap with your real moneymakers — `seo-article-writer`, `merlino-magic-blog`, `paa-content-writer`, `lowfruit-article-writer`, `thanos-optimizer`. - **No SEO strategy**: no keyword research, no intent analysis, no PAA mining, no SERP analysis, no E-E-A-T, no entity SEO, no local SEO. That's your entire business and it's missing. **The one thing worth lifting**: the hardened IndexNow Edge Function pattern (verifyJwt + host validation). That's a ~60-line snippet, not a skill install. I can drop it as a Next.js route handler into any project in 5 minutes if you ever want instant Bing/Yandex indexing on content drops. **Verdict**: ignore the skill. Sticklight is a Vite/Bolt-clone builder pitching skills to its own users. You're not their target.

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>Waiting.</result> <usage><total_tokens>90760</total_tokens><tool_uses>0</tool_uses><duration_ms>2363</duration_ms></usage> </task-notification> [ASSISTANT] Standing by.

[USER] <<autonomous-loop-dynamic>> [ASSISTANT] Done. Full delivery report: ## DELIVERED: Feedly API SOP site **Site:** https://feedly-api-docs.vercel.app **Repo:** https://github.com/mmerlino23/feedly-api-docs (private) ### Pages - 77 total: 7 Guides, 65 API Reference, 5 Changelog - 13 sidebar groups: Getting Started, Articles & Search, Enterprise & Boards, AI Feeds & Ask AI, Threat Landscape, Cyberattacks Agent, CVEs, TTP Agent, Threat Actors, IoCs, Malware, Emerging Trends, Company Insights, Webhooks - All external links stripped (zero outbound links rule honored) ### RAG chat (verified working) - OpenAI Responses API + File Search - Vector store: `vs_6a0b021268548191ba721ce47be8ec0f` (81 files indexed) - Verified: "How do I authenticate?" → returned correct answer + curl example, cited `authorization.md` - Verified: "What endpoints for CVE data?" → returned 3 endpoints with correct paths, cited `cve-insights-card.md` ### Vaults (both) - Canonical: `D:/Ecosystem/vaults/feedly-api-docs/` — 81 md files - Legacy: `E:/Merlino Vault/Resources/SOP-Sites/feedly-api-docs/` — 81 md files ### Memory ingestion - Master Brain: scope `sop:feedly-api-sop`, 509 chunks (text-embedding-3-small) - merlino-sops MCP: 22nd category `feedly` (agent: Raven), tsc passes - sop-sites-registry.json: appended (4 sites total) - Memz/Mem0/Hindsight: fanned out via agent-memory.py ### Build notes - **Firecrawl out of credits** → pivoted to Readme.io `llms.txt` URL inventory + direct curl (Readme.io serves clean `.md` per page) - **Bug caught + fixed:** `echo $KEY | vercel env add` adds trailing `\n` → OpenAI SDK threw "Connection error" via `TypeError: ... not a legal HTTP header value`. Switched to `printf "%s"`. Saved as feedback memory for future SOP builds. - Increased function timeout to 60s (Responses API + file_search runs ~10-13s) - Enabled `cleanUrls: true` ### Restart required Restart Claude Code so the merlino-sops MCP picks up the new `feedly` category. Test query: ask any session about Feedly Ask AI endpoints and confirm scope `sop:feedly-api-sop` returns hits. **VERIFICATION:** FULL (production smoke-tested home + reference + RAG chat with 2 queries, citations returned) **VAULT:** 81 files copied to both D and E vault paths **MEMORY:** scope=sop:feedly-api-sop chunks=509 memz=ok mcp-category-added=yes registry-updated=yes restart-required=yes

[USER] [tool_result] 100 - [Process Catalog v0](project_process-catalog-v0.md): Supabase table in Master Brain. 7,229 rows post-dedup. 489 real skills, 2 executable-process. Replaces "skills" as agent-queryable surface. 101 - [Petey Agent](project_petey-agent.md): Process compiler subagent with persistent memory. Reads raw knowledge, emits structured workflows/SOPs/catalog rows/agent defs. Smoke test: local-seo-site-builder, 48 steps extracted. 102 - [Skill Dedup 2026-05-16](project_skill-dedup-2026-05-16.md): Purge [ASSISTANT] Done. Full delivery report: ## DELIVERED: Feedly API SOP site **Site:** https://feedly-api-docs.vercel.app **Repo:** https://github.com/mmerlino23/feedly-api-docs (private) ### Pages - 77 total: 7 Guides, 65 API Reference, 5 Changelog - 13 sidebar groups: Getting Started, Articles & Search, Enterprise & Boards, AI Feeds & Ask AI, Threat Landscape, Cyberattacks Agent, CVEs, TTP Agent, Threat Actors, IoCs, Malware, Emerging Trends, Company Insights, Webhooks - All external links stripped (zero outbound links rule honored) ### RAG chat (verified working) - OpenAI Responses API + File Search - Vector store: `vs_6a0b021268548191ba721ce47be8ec0f` (81 files indexed) - Verified: "How do I authenticate?" → returned correct answer + curl example, cited `authorization.md` - Verified: "What endpoints for CVE data?" → returned 3 endpoints with correct paths, cited `cve-insights-card.md` ### Vaults (both) - Canonical: `D:/Ecosystem/vaults/feedly-api-docs/` — 81 md files - Legacy: `E:/Merlino Vault/Resources/SOP-Sites/feedly-api-docs/` — 81 md files ### Memory ingestion - Master Brain: scope `sop:feedly-api-sop`, 509 chunks (text-embedding-3-small) - merlino-sops MCP: 22nd category `feedly` (agent: Raven), tsc passes - sop-sites-registry.json: appended (4 sites total) - Memz/Mem0/Hindsight: fanned out via agent-memory.py ### Build notes - **Firecrawl out of credits** → pivoted to Readme.io `llms.txt` URL inventory + direct curl (Readme.io serves clean `.md` per page) - **Bug caught + fixed:** `echo $KEY | vercel env add` adds trailing `\n` → OpenAI SDK threw "Connection error" via `TypeError: ... not a legal HTTP header value`. Switched to `printf "%s"`. Saved as feedback memory for future SOP builds. - Increased function timeout to 60s (Responses API + file_search runs ~10-13s) - Enabled `cleanUrls: true` ### Restart required Restart Claude Code so the merlino-sops MCP picks up the new `feedly` category. Test query: ask any session about Feedly Ask AI endpoints and confirm scope `sop:feedly-api-sop` returns hits. **VERIFICATION:** FULL (production smoke-tested home + reference + RAG chat with 2 queries, citations returned) **VAULT:** 81 files copied to both D and E vault paths **MEMORY:** scope=sop:feedly-api-sop chunks=509 memz=ok mcp-category-added=yes registry-updated=yes restart-required=yes

[USER] [tool_result] package.json scripts: ['start', 'ingest', 'smoke'] Shell cwd was reset to D:\ClaudeDev [ASSISTANT] [tool_use:Bash]

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>15G. Waiting.</result> <usage><total_tokens>90617</total_tokens><tool_uses>0</tool_uses><duration_ms>3146</duration_ms></usage> </task-notification> [ASSISTANT] Standing by for your decision on speed/scope.

[USER] ok yes but u watn to see a tet ouput of tha tactually seo artucle which is the real point [ASSISTANT] Honest take: **don't fragment your stack just to have variety**. Variety has a real cost for you specifically. **What ShipFast uses (Marc Lou's stack):** - Next.js (App Router) + TypeScript - Tailwind + **DaisyUI** (not ShadCN) - MongoDB or Supabase (two versions sold) - NextAuth (Google + Magic Link) - Stripe + Lemon Squeezy - Resend for email - Vercel hosting So ShipFast is basically your stack minus ShadCN, plus DaisyUI. Marc bet on Next.js for the same reason you should: SEO out of the box. **Why Vite + React is a downgrade for YOU specifically:** - Your entire business is local SEO. SPAs make SEO harder, which is exactly why that Sticklight skill needs 14 sections of `@unhead/react`, `data-hid` fallbacks, static sitemap hacks. Next.js gives you all of that natively with `generateMetadata()` and route handlers. - You'd lose Server Components, ISR, native sitemap routes, and the Vercel/Next pairing. - You'd fork your skill ecosystem (488 skills tuned for Next.js patterns). **When Vite IS the right call (legit secondary stack):** - Browser extensions (Chrome/Edge) - Embedded widgets you drop into client sites - Tauri/Electron desktop apps (HawkeyePanel desktop version?) - Pure internal tools behind auth where dev speed > SSR - Learning/prototyping new React libs fast **Recommendation:** Keep Next.js as your one primary stack. Add Vite **only** when you build a Chrome extension or a Tauri desktop wrapper. Don't run two stacks in parallel for SEO-facing work, you'll dilute your template library and CLI tooling. Want me to build that hardened IndexNow piece as a Next.js route handler so it slots into any of your current Vercel projects?

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>13G. Waiting.</result> <usage><total_tokens>90476</total_tokens><tool_uses>0</tool_uses><duration_ms>2386</duration_ms></usage> </task-notification> [ASSISTANT] Standing by for your decision on speed/scope.

[USER] how many pages is this? does it cover everything? do they have API or MCP? nothing shows in the chat or stie - t seems that the documentation files you uploaded do not contain information about using the API in Divi. However, I can provide a general overview of how to use APIs in Divi. ### Using the JavaScript API in Divi 1. Initialization: To use the Divi JavaScript API, ensure that your custom scripts are loading after the Divi scripts. You can do this by enqueueing your script correctly in WordPress. ``php function my_custom_script_enqueue() { wp_enqueue_script( 'my-custom-script', get_template_directory_uri() . '/js/my-custom-script.js', array('jquery'), null, true ); } add_action( 'wp_enqueue_scripts', 'my_custom_script_enqueue' ); ` 2. Accessing the API: The Divi JavaScript API is available globally. You can access it via the et object. For example, to trigger a modal in Divi: `javascript jQuery(document).ready(function($) { et.modules.modal.open('my-modal-id'); }); ` 3. Handling events: The API allows you to hook into various events. You can listen for changes or actions and respond accordingly. `javascript et.modules.form.on('submit', function(event) { // Handle form submission }); `` 4. Extending functionality: You can extend or customize Divi's existing modules by writing your own JavaScript functions that leverage the API capabilities. If you need more specific details or functionality examples, please let me know! [ASSISTANT] All shipped. Final delivery report: ``` DELIVERED: VitePress SOP site for Divi Developer Documentation, 25 pages with RAG chat grounded on every page. PROOF: - Site: https://elegant-themes-developers-docs.vercel.app - GitHub: https://github.com/mmerlino23/elegant-themes-developers-docs (private) - Vault: D:/Ecosystem/vaults/elegant-themes-developers-docs/ (25 md files) - RAG chat: Vector store vs_6a0affbb09148191a6826d1159f0c8a2 (22 files uploaded) Live test: "How do I register a custom module..." returned correct File Search answer with citation to extension-build-pipeline.md - Pages live: home, sops/, modules/, hooks/, code-reference/, workflows/, quick-ref/ — all 10 spot-checked URLs return HTTP 200 VERIFICATION: FULL - lint/build: VitePress build complete in 5.14s (no errors) - routing: Vercel cleanUrls=true verified across 10 content pages - api: /api/chat returns 200 with valid response + file citations - content: spot-checked 5 pages with real body text, code blocks, tables - links: zero external http links in any .md (grep verified) VAULT: 25 files copied to D:/Ecosystem/vaults/elegant-themes-developers-docs/ MEMORY: scope=sop:elegant-themes-developers-docs chunks=77 memz=ok (Memz + Mem0 + Hindsight fan-out via agent-memory.py) mcp-category-added=yes (divi, 20th category) registry-updated=yes (D:/Ecosystem/sop-sites-registry.json) restart-required=yes ``` **Sources used:** Firecrawl ran out of credits after the 1-page landing crawl, so I fell back to direct curl + the public WP REST endpoint (`/wp-json/wp/v2/posts?categories=22`) which returned all 15 child posts with full rendered HTML. Converted via markdownify, stripped all outbound links per SOP rule. **Two bugs caught and fixed in-flight:** 1. OpenAI Node SDK threw `Connection error` on Vercel — swapped to raw `fetch` against `api.openai.com/v1/responses` (matches the documented `reference_openai-sdk-vercel-bug` memory). 2. VitePress emits `.html` files but Vercel needed `cleanUrls: true` to serve `/foo` from `/foo.html`. **Next:** restart Claude Code so the `merlino-sops` MCP picks up the new `divi` category, then test a query like *"What are the required parameters in get_fields() for a Divi module?"* against that scope.

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>12G. Waiting.</result> <usage><total_tokens>90333</total_tokens><tool_uses>0</tool_uses><duration_ms>2569</duration_ms></usage> </task-notification> [ASSISTANT] Real talk on timing: **Current pace:** ~11 GB copied in ~22 min = ~30 GB/hr **Sizes left to copy (estimates, exact still computing):** - Merlino Vault: ~30-50 GB remaining (about 25% done at 11 GB) - Master-Vault-Backup: 5-15 GB - Master-Vault-RAG-Archive: 50-100 GB (RAG archives are heavy) - **Master-Vault-Media-Archive: unknown, likely 100-300 GB** (the wildcard — could double the total) - SEO ROCKSTARS: unknown **Honest range:** - **Best case (no media archive bloat): 4-6 hours** - **Realistic: 8-12 hours** - **Worst case (huge media): 15-20 hours** Your own memory just flagged: *"Long-running agents (12+ hours) are unreliable and should be short-lived and task-focused."* That tracks. If Master-Vault-Media-Archive turns out to be 200+ GB, the current pattern is risky. **Options to speed up / de-risk:** 1. **Skip Master-Vault-Media-Archive** (only one with images/video, biggest unknown) — likely cuts total by 50-70% 2. **Filter to .md/.txt/.json only** on all remaining vaults — drops media bloat, finishes in 1-3 hours 3. **Let it run as-is** — accept the 8-12 hour estimate, robocopy is resilient and the dest is being verified per-file What do you want?

[USER] how long will this take [ASSISTANT] Standing by.

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>11G copied total. Pace picked up. Waiting.</result> <usage><total_tokens>90181</total_tokens><tool_uses>0</tool_uses><duration_ms>2713</duration_ms></usage> </task-notification> [ASSISTANT] Standing by.

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>Waiting.</result> <usage><total_tokens>90044</total_tokens><tool_uses>0</tool_uses><duration_ms>3430</duration_ms></usage> </task-notification> [ASSISTANT] Standing by.

[USER] [tool_result]  The language 'env' is not loaded, falling back to 'txt' for syntax highlighting.  The language 'env' is not loaded, falling back to 'txt' for syntax highlighting.  The language 'env' is not loaded, falling back to 'txt' for syntax highlighting.  The language 'env' is not loaded, falling back to 'txt' for syntax highlighting.  The language 'env' is not loaded, falling back to 'txt' for syntax highlighting.  The language 'env' is not loaded, [ASSISTANT] [tool_use:Bash]

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>6G total. Steady pace ~1G/2min. Waiting.</result> <usage><total_tokens>89887</total_tokens><tool_uses>0</tool_uses><duration_ms>2671</duration_ms></usage> </task-notification> [ASSISTANT] Standing by.

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>5G copied. Waiting.</result> <usage><total_tokens>89743</total_tokens><tool_uses>0</tool_uses><duration_ms>2836</duration_ms></usage> </task-notification> [ASSISTANT] Standing by.

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>Confirmed dest is growing: 6,605 .md files and 3.2G in merlino-vault. Robocopy working well. Waiting for monitor events.</result> <usage><total_tokens>90670</total_tokens><tool_uses>1</tool_uses><duration_ms>10525</duration_ms></usage> </task-notification> [ASSISTANT] Standing by.

[USER] <task-notification> <task-id>bks42remd</task-id> <tool-use-id>toolu_018TDLPkhRvrNtbuRYRedwSa</tool-use-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\bks42remd.output</output-file> <status>completed</status> <summary>Background command "Check log + merlino-vault progress" completed (exit code 0)</summary> </task-notification> [ASSISTANT] Standing by.

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>Progressing normally. 4G copied total. Waiting.</result> <usage><total_tokens>89088</total_tokens><tool_uses>0</tool_uses><duration_ms>3270</duration_ms></usage> </task-notification> [ASSISTANT] Standing by for the wakeup.

[USER] [tool_result] 85 - [ByteRover Docs SOP + Agent Skill](project_byterover-docs-site.md) — Built 2026-05-14: 70 pages via Mintlify llms.txt fetch, live at https://byterover-docs.vercel.app, agent skill byterover-cli at merlino-skills/ 86 - [Stop Asking Obvious Questions](feedback_stop-asking-obvious-questions.md) — When Mike sends URLs + says 'get data' — EXECUTE. Don't ask 'is it public?' preemptively. 87 - [Memory Ecosystem 8 Stores](project_memory-ecosystem-8-stores.md) — 2026-05-14: agent-memory.py fans out [ASSISTANT] [tool_use:Bash]

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>Steady progress. Waiting.</result> <usage><total_tokens>88943</total_tokens><tool_uses>0</tool_uses><duration_ms>2463</duration_ms></usage> </task-notification> [ASSISTANT] Standing by for the wakeup.

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>Progress continuing: 3.5MB log, 815G free (3G copied total). Waiting.</result> <usage><total_tokens>88781</total_tokens><tool_uses>0</tool_uses><duration_ms>2897</duration_ms></usage> </task-notification> [ASSISTANT] Wakeup set for 30 min. Standing by.

[USER] <task-notification> <task-id>ab9b2ab6c7f5494b6</task-id> <tool-use-id>toolu_01UGjpv3KKFseNQgBy5tmKqj</tool-use-id> <output-file>C:\Users\mikem\AppData\Local\Temp\claude\D--ClaudeDev\a05a7db0-fdbd-4d5a-9019-b037f8cf7225\tasks\ab9b2ab6c7f5494b6.output</output-file> <status>completed</status> <summary>Agent "Robocopy E: vaults to F:\HERE" completed</summary> <result>Waiting for notification.</result> <usage><total_tokens>91837</total_tokens><tool_uses>40</tool_uses><duration_ms>1211549</duration_ms></usage> </task-notification> [ASSISTANT] [tool_use:ScheduleWakeup]

[USER] can adjut to work with my set uo but i am NOT lcoked into my tech stack ata ll why dont I use reacht? and vite? I should have more than one tech stack so I want to use both and what does shipfast use [ASSISTANT] Quick take: - **What it is**: Sticklight is a skill marketplace tied to their own Vite SPA builder. The skill is decent but stack-locked to **React + Vite + @unhead/react + Supabase Edge Functions**. Doesn't map to your Next.js stack since the App Router has built-in `metadata` / `generateMetadata` and dynamic sitemap routes natively. - **What's actually worth lifting**: the **IndexNow Edge Function** pattern with `verifyJwt: true` + strict host validation. That's the one piece that's genuinely good and stack-agnostic. Open-relay protection on IndexNow is rarely documented this cleanly. Could fold it into your `technical-seo-agent` or `schema-generator` skill as a Next.js route handler version. - **Skip the rest**: `@unhead/react`, `useSEO` hook, static `data-hid` fallbacks, manual sitemap.xml — all redundant for Next.js. Your existing skills (einstein, technical-seo-agent, schema-spider, auto-schema-from-template, llm-mentions) already cover semantic HTML, schema, sitemaps, robots, OG tags, Core Web Vitals. Want me to spin out a small `indexnow-integration` skill that wraps the hardened pattern as a Next.js Route Handler (instead of Supabase Edge Function), so it drops into any of your Vercel projects?