famdeck

You are the Session Intelligence analyst. You receive structured signal data from Claude Code sessions and produce natural-language findings for issue filing.

Assess a project's verification infrastructure and assign an autonomy readiness level (L0-L2). Use when the user asks 'is this project ready for autonomous work', 'check autonomy level', 'what verification tools are set up', 'assess project readiness', or when starting work on a new project to understand its quality baseline. Pairs with /autonomy-scaffold to fill gaps.

Generate missing verification tool configs (linter, test runner, CI pipeline, coverage) to bring a project up to autonomy level L2. Use when the user says 'scaffold quality tools', 'add linting', 'set up CI', 'improve test infrastructure', or after /autonomy-assess shows the project is below L2. Shows a dry-run plan before applying.

Autonomous development loop: reconcile orphan Beads issues into BMAD sprint planning, execute stories, close completed issues, repeat until done. Use when the user says 'autopilot', 'start working autonomously', 'run the dev loop', 'work through the backlog', or wants hands-off autonomous development using Beads + BMAD.

Invoke this skill whenever the user has finished an implementation task and needs an outside check that it actually works and meets requirements. This covers: post-implementation sanity checks ("is it done?", "does it pass?", "am I good?"), pre-ship quality gates (lint + types + tests + AC check), acceptance criteria audits against story specs or beads issues, and explicit /verify commands. The distinguishing intent: the user is not asking for new code — they want judgment on work already done. This skill runs regression tests, evaluates each acceptance criterion with evidence, and writes tests to lock in new behavior. Do NOT use when the user wants to implement something new, refactor code, or run a single named test command they specified.

4 hooks across 4 events