setup-devinat1-skills

# Setup devinat1 Skills Scaffold the per-repo configuration that these skills assume: - **Agent memory** — stable project ID for learning skills to scope MCP memories - **Blog directory** — where `/blog` and `/update-blog-refs` read existing posts - **MCP integrations** — which optional MCP servers you use (agentmemory, Granola, Todoist) This is a prompt-driven skill, not a deterministic script. Explore, present what you found, confirm with the user, then write. ## Process ### 1. Explore Look at the current repo to understand its starting state: - `git remote -v` — identify the repo and derive a default project slug (`owner-repo`) - `AGENTS.md` and `CLAUDE.md` at the repo root — does either exist? Is there already an `## Agent skills` section? - `docs/agents/` — does prior setup output already exist? ### 2. Present findings and ask Summarise what's present and what's missing. Walk the user through three decisions **one at a time**. **Section A — Agent memory project ID.** > Explainer: The learning skills recall prior knowledge for routing and personalization. `/dunning-krueger`, `/break-it`, and `/experience` also write learning evidence to the `user-agentmemory` MCP server. They need a stable project slug to scope memories. Default: `owner-repo` from `git remote get-url origin` (e.g. `devinat1-claude`). Ask if they want to override the project ID. **Section B — Blog content directory.** > Explainer: `/blog` and `/update-blog-refs` read your existing posts to match voice and suggest cross-links. Default: ask the user for their blog content directory (no default assumed). **Section C — MCP integrations.** > Explainer: Some skills use MCP servers. Record which you have so skills know what's available. Ask about agentmemory (learning skills and experience), Granola (meeting-feedback, momtest), and Todoist (focus, ramble). Options: configured / not configured / not needed. ### 3. Confirm and edit Show drafts of: - The `## Agent skills` block for `CLAUDE.md` or `AGENTS.md` - `docs/agents/agent-memory.md`, `docs/agents/blog-directory.md`, `docs/agents/mcp-integrations.md` Let the user edit before writing. ### 4. Write **Pick the file to edit:** - If `CLAUDE.md` exists, edit it. - Else if `AGENTS.md` exists, edit it. - If neither exists, ask the user which one to create. If an `## Agent skills` block already exists, update it in-place — don't append a duplicate. The block: ```markdown ## Agent skills ### Agent memory [one-line summary of project ID]. See `docs/agents/agent-memory.md`. ### Blog directory [one-line summary of blog path]. See `docs/agents/blog-directory.md`. ### MCP integrations [one-line summary]. See `docs/agents/mcp-integrations.md`. ``` Write `docs/agents/*.md` using the seed templates in this skill folder as starting points: - [agent-memory.md](./agent-memory.md) - [blog-directory.md](./blog-directory.md) - [mcp-integrations.md](./mcp-integrations.md) Substitute the user's chosen values into the written files. ### 5. Done Tell the user setup is complete and which skills now read from `docs/agents/*.md`. They can edit those files directly later. Remind them that agentmemory requires `npx @agentmemory/agentmemory@0.9.27` running at `AGENTMEMORY_URL`.