Skills

Append JSONL interaction records to ~/.ai/audit/interactions/YYYY-MM.jsonl for every tool use, session start/end, and stop event.

Log Bash commands and their arguments before execution.

Block direct commits, pushes, and merges to protected branches (main, master).

Write a CHECKPOINT.md snapshot on each turn to support session recovery.

Block destructive gh CLI operations (repo delete, branch force-push, etc.).

Block destructive kubectl operations (delete, drain, cordon) in production contexts.

Block destructive terraform operations (destroy, force-replace) without confirmation.

Strip --no-verify flags from git commands to enforce pre-commit hook execution.

Redact 1Password secret values from tool output before they appear in the conversation.

JSON pattern library used by secret-block and secret-precommit for secret detection.

Block Bash commands that would print secret values (env, printenv, cat .env, etc.).

Git pre-commit hook that scans staged files for secrets before they are committed.

Enforce §U17 worktree placement rules — worktrees must live in canonical locations.

Navigate the convergent-systems atom ecosystem to find, fetch, or scaffold atoms. Routes 'apply brand X' to brand-atoms, 'use persona Y' to persona-atoms, 'load skill Z' to skill-atoms, etc. Covers discovery, cross-atom reference resolution, and new-atom scaffolding.

Snapshot in-flight work — goals, decisions, file paths, branch state, next step — to a structured checkpoint file so a later session can resume exactly where this one left off.

Remove dead code, unused imports, commented-out blocks, and resolved TODOs. Behavior-preserving: no logic changes, only deletion of unreachable or inert code.

Generate a conventional-commit message from the staged diff and commit. Reads the diff, analyzes the change, picks the right Conventional Commits prefix (feat/fix/refactor/chore/docs/test/ci), and commits with a descriptive message.

Generate Mermaid diagrams (flowchart / sequence / state / class / ER / gantt) from code or systems.

Read-only code explainer with audience-tuned depth (non-technical / junior / practitioner / senior) and path:line citations. Never modifies files.

Unified make dispatcher. Delegates to the correct sub-skill (work/clean/build/test) based on the subcommand argument.

Discover, fetch, or scaffold skill atoms via the atoms-discover skill. Routes to the atoms binary for ecosystem operations.

Commit staged changes, push the current branch, open a pull request, and merge after CI passes.

Two-command git workspace lifecycle manager: purge merged branches/worktrees/stashes, aggregate remaining in-flight work into one branch, then hand off to make build.

Health check for the AI governance stack and per-repo CI: runs 'ai doctor', checks hook permissions, verifies CLAUDE.md constitution includes, and reports CI status on open PRs.

Build a GitHub Project (Projects v2) with a three-level Epic → Feature → Task hierarchy, sourced from GOALS.md, existing GitHub issues, and Azure DevOps work items. Uses native GitHub sub-issues to wire the hierarchy. Confirms the derived tree before creating anything.

Adversarial code reviewer sweep on the current branch diff against main: stubs, placeholders, dead imports, unwired exports, and TODO/FIXME in production paths.

Activate PM discipline and run the multi-agent delivery team in one shot. Recovers from checkpoints, ingests the issue queue, partitions tasks, spawns Tech Lead agents per domain, runs TDD pipelines, code reviews, and builds.

Sprint status snapshot: open PRs and their CI state, stale branches, sync gap between local and origin, and any uncommitted work.

Pull all repos in the workspace (primary + submodules) to their default branch heads.

Discover and run the project test suite across all workspace modules, report failures with file:line citations, and optionally fix them.

Generate an ONBOARDING.md for the current repo covering: stack, prerequisites, setup, run, test, structure, and common commands. Targeted at a new contributor picking up the repo cold.

Open a GitHub pull request with a body generated from the branch's commits. Derives title and summary from commit history.

Set up AI workspace config for this project. Installs project-claude (Claude Code) and project-copilot (GitHub Copilot) together.

Set up and maintain AI workspace configuration for Claude Code: project.yaml, .claude/settings.json, and CLAUDE.md. Use when initialising a project for Claude or editing its Claude-specific config.

Set up and maintain AI workspace configuration for GitHub Copilot CLI: project.yaml and .github/copilot-instructions.md. Use when initialising a project for Copilot or editing its Copilot-specific config.

[Retired — use /project-claude or /project-copilot instead] Manage AI workspace configuration for a project — project.yaml, integration files, and tool settings.

Behavior-preserving refactor with test gates and the bug-report-separately rule. Tests must be green before and after. Bugs found during refactor are reported separately, not fixed inline.

Repository structure tool — monorepo or polyrepo migration in either direction, preserving git history.

Mid-session blocked-state report. Reports ready state or lists active waits (biometric, pr-merge, scope-decision, out-of-band, background) using the Common.md §U7 typed wait-reason format.

Run repo tests via project.yaml's tooling.test_command and report failures with path:line citations.

Configure the Claude Code harness via settings.json. Handles: automated behaviors (hooks), permissions, env vars, hook troubleshooting, and changes to settings.json/settings.local.json files.