fabienjuif
- 6 skills
- 0 followers
- 12 hours ago last updated
- ▌ Idea · fabienjuifQuickly capture an idea about the current project into docs/IDEA.md — a "what's next" backlog. Ideas can be anything project-related, technical or not, horizontal (tooling, process, docs) or vertical (a feature). One dated entry with just enough context that a future session or human can pick it up cold. Also handles pruning ("/idea unbloat" — interactive curation where surviving ideas earn a survived ×N marker) and prioritizing ("/idea sort" — an interactive triage session where the USER decides priority buckets, seeded by the survival counter, and the file is rewritten in that order). Use when the user says "/idea", "capture this idea", "note this idea", "remember this for later", "add to the idea backlog", "/idea unbloat", "prune the ideas", "/idea sort", "help me prioritize the ideas", or drops a "what if / we could / someday / next we should" thought they don't want to act on right now. For work to start now, use make-plan instead.
- ▌ Catch Up · fabienjuifPersonal daily catch-up dashboard — gathers, for the configured sources, the Slack messages where you're involved in the last week plus your Slack "save for later" items, the Jira tickets assigned to you or open & unassigned, and the open PRs in your repos (open, or where you're mentioned), then distills them into one prioritized "do these first" list rather than dumping everything. On first run, if there's no config, it walks you through setup. Use when asked to "catch up", "what's on me", "my standup", "what needs my attention", "/catch-up", or any "what did I miss / what's waiting on me" request.
- ▌ Make Plan · fabienjuifProduce a plan for a non-trivial task as a self-contained docs/plans/<name>/ directory (CLAUDE.md, README.md, ROADMAP.md, PROMPT.md, changelogs/) so a fresh agent session AND a human can pick it up later. Use whenever asked to "create a plan", "plan this out", or before implementing anything multi-step or architectural.
- ▌ Pr Review · fabienjuifReview a pull request and post ONE consolidated review comment that updates in place across re-reviews — each pass collapses the prior review into a `<details>` block instead of stacking new comments, so the thread stays clean and diffable. This is the reason to reach for it over `/code-review --comment` (which posts fresh inline threads every time). Wraps the built-in /code-review for analysis; adds modes (local print vs push), continuity (re-reads the last review), and history collapse. Use when asked to "review this PR", "/pr-review", "/pr-review push", "/pr-review local", "post a review", or "re-review after my changes".
- ▌ Capture Docs · fabienjuifAt the end of a session, capture what was learned into the repo's documentation so it isn't lost and future sessions start cheaper. Use when the user says "capture docs", "save what we learned", "update docs", "sync docs", or runs "/capture-docs", typically when wrapping up. Relies on the conversation already in context plus cheap git signals — it does NOT re-research the codebase. Updates human-facing docs (READMEs, docs/) and AI-facing docs (CLAUDE.md, AGENTS.md), keeping AI docs as pointers to human docs, and ensures what it writes is discoverable from the nearest CLAUDE.md/AGENTS.md.
- ▌ Fix Stale Doc · fabienjuifFull-repo documentation sweep. Fans out one sub-agent per module / docs subdir / README cluster to (1) fix stale claims that no longer match the code, (2) remove duplication, and (3) compact verbose prose. Use when the user runs "/fix-stale-doc", asks to "fix stale docs", or "clean up documentation".