CrossCheck

Process autonomous task queue from do-work/ folder

Enter plan mode for complex tasks (pour energy into the plan for 1-shot implementation)

Initiate autonomous PR review process with Codex agent

Run comprehensive repo assessment with Codex (every 3 PRs)

Comprehensive security audit covering 10 threat categories

Install all automation for a new repo (git hooks + GitHub Actions)

Automated PR submission - runs pre-checks, creates PR, and starts review

Update CrossCheck to the latest version — pulls from main. Skills and agents update instantly via symlinks; new ones get linked automatically.

Extract and persist a non-obvious discovery from the current session as a reusable skill. Use after: debugging a non-obvious error, finding a workaround through trial-and-error, resolving an environment/config/dependency issue the docs didn't cover, discovering a project-specific pattern through investigation, or completing any task where the solution required genuine discovery rather than straightforward documentation lookup.

Automated Playwright-based web app testing. Screenshot, inspect DOM, discover selectors, interact with elements, and run visual regression checks against a running localhost app. Use when: "test webapp", "browser test", "screenshot app", "check UI", "visual regression", "end-to-end test", or verifying a web app works after changes.

Systematic audit for common failure modes and bugs

Property-based and adversarial input testing — generates random/adversarial inputs to find crashes, hangs, and invariant violations

Verify test suite effectiveness — introduces small code mutations and checks if tests catch them. Surviving mutants reveal weak tests.

Active exploit verification — writes and runs throwaway exploit tests to confirm or disprove security findings

Install and configure Claude Code plugins alongside CrossCheck skills

Per-repo behavioral memory stored in .codex/napkin.md. Use to initialize or update the napkin for any project. Tracks repo-specific mistakes, user corrections, and surprising gotchas that recur across sessions. Read silently at session start if the file exists. Write during work. Curate aggressively. Complement to MEMORY.md (which is cross-project and user-scoped); napkin is repo-scoped and committable.

Track AI coding tool token usage, costs, and environmental impact

Batch git branch cleanup - generates single approval script

Remove merged or abandoned git worktrees

Delegate a coding task to OpenAI Codex agent for autonomous execution

Atomic Git Commit

Create new git worktree for parallel development

Add or update datetime timestamps in documentation files

Get multi-model opinions on a problem (Codex + Gemini + Critic)

Manage files in /garbage folder for deletion

Delegate a coding task to Google Gemini agent for autonomous execution

List all active git worktrees

Comprehensive pre-PR checklist - actually runs checks

Customize Codex statusline for better context awareness

Find and eliminate technical debt (run at end of every session)

Run all git cleanup operations in sequence: worktrees, branches, and stashes. Use before or after repo assessment, at the end of a sprint, or whenever local git state has accumulated clutter. Chains /cleanup-worktrees, /cleanup-branches, and /cleanup-stashes with a combined summary at the end.

Review and selectively drop git stashes. Lists all stashes with branch, age, and file summary. Flags stashes whose branch no longer exists (orphaned work) or has been merged to main (likely superseded), and stashes older than 14 days (likely forgotten). Generates a drop script for user approval — never drops stashes directly.