claude-config

Draft, polish, and create a well-formed GitHub issue with required structure

Post a structured checkpoint comment to the relevant issue and commit current code state

Close out issue work - ensure PR exists, run the full CodeRabbit + /code-review gate, merge, close all issues in the working set, clean branches, and update PROGRESS.md

Fresh session pickup using the most recent checkpoint comment and git log — reconstructs working context for active issue work

Recommend juggling pairs, user decides, create correctly-named branch for issue work

Update the active issue with a design decision and cascade impact to all open issues and PRDs

Complete PRD implementation workflow - create branch, push changes, create PR, merge, and close issue

Update PRD progress based on git commits and code changes, enhanced by conversation context

Create documentation-first PRDs that guide development through user-facing content

Pre-PR verification loop that runs build, type check, lint, tests, and security checks. Use before creating pull requests or when you want to validate code quality.

Create Anki cards from the current conversation. Invoke when learning a concept and want to capture it for spaced repetition.

Create and save Anki cards autonomously without approval. Use when cards should just get made.

Resume work from a previous session. Reads PRD state, PROGRESS.md, git state, task list, and layered journal context to summarize where work left off and suggest the next step. Use at the start of a work session. For mid-session context recovery after /compact, use /post-compact instead.

Show Claude Code token usage and cost broken down by repo and model. Use when checking session spend, which repo costs the most, cache efficiency, or total cost over a period. Triggers on "how much did that cost", "token usage", "cost breakdown", "cache hit rate", "which repo is most expensive", "billing", or "how much am I spending".

Re-anchor context after a compaction event. Reads CLAUDE.md, the active PRD, and git state, then reports orientation. Use immediately after /compact to recover lost context.

Fetch all open GitHub issues — PRDs and standalone issues — for project planning

Write and review any document an AI will read and act on — system prompts, SKILL.md/CLAUDE.md/rules files, PRDs, GitHub issues, agent specs. If an AI reads it and acts on it, it's a prompt.

Write validated documentation by executing real commands and capturing actual output. Use this skill whenever documentation writing is needed — from PRD milestones, CLAUDE.md instructions, or user requests.

Research a topic, technology, or question using web search and documentation. Use this skill before adopting new technologies or when current documentation is needed.

Update PRD based on design decisions and strategic changes made during conversations

Close a PRD that is already implemented or no longer needed

Analyze existing PRD to identify and recommend the single highest-priority task to work on next

Start working on a PRD implementation

Enable autonomous PRD mode for the current project. Installs YOLO skill symlinks, SessionStart hooks, and permissions.

Disable autonomous PRD mode — swap YOLO skill symlinks to careful variants, remove hooks and permissions