claude-settings

Create logically grouped, atomic git commits with well-formatted commit messages following best practices. Use this skill when you need to commit changes to a git repository with proper message formatting and atomic grouping. Before creating commits, reflect on the work just completed and surface any open questions or unresolved decisions (judgment calls made without checking, deferred edge cases, defaults picked when multiple options were viable, trade-offs that could reasonably go the other way) via AskUserQuestion so they get discussed with the user rather than silently committed.

Review code with grug-brain anti-complexity philosophy. Use when reviewing PRs, diffs, or code changes to identify unnecessary complexity.

Implement an epic's beads in a worktree with logical commits, multi-agent review, fix pipeline, and PR description generation. Use when you want to implement this epic, burn down the beads, execute the plan, or start working on an epic.

Comprehensive branch review using parallel background agents with Codex validation. Each reviewer runs as an independent background task - results are collected only after all agents complete, eliminating premature synthesis. Use for thorough pre-PR review of any branch size.

Implement fixes for code review findings using a parallel agent team. Accepts a review report (from /team-branch-review, /team-commit-review, or pasted), interviews the user on which findings to fix (or runs autonomously with --auto), then spawns agents to implement fixes in parallel with Codex validation.

Comprehensive branch review using a parallel agent team with Codex validation. Use when reviewing all commits on a branch before creating a PR, especially for large or complex branches that benefit from multi-perspective review. Trigger on "review my branch", "check before I PR", "full code review", "pre-PR review", "team review", or "review before merge".

Comprehensive commit-scoped review using a parallel agent team with Codex validation. Use when reviewing specific commits rather than an entire branch - ideal for auditing a single commit, the last N commits, or a specific SHA before or after merging. Trigger on "review this commit", "review last 3 commits", "review HEAD", "commit review", or "check this commit".

Plan issues using autonomous AI debate between Claude (opus) and Codex. Heavyweight and thorough - best for complex architectural work, large refactors, or ambiguous problems that need deep analysis. Use when the work is too important for the lightweight issue-plan or when you need maximum rigor in discovery and planning without user involvement.

Plan issues using autonomous AI debate between Claude (haiku) and Codex. Lightweight and fast - best for routine multi-step work where full user involvement or heavy models are unnecessary. Use when planning standard features, bug fixes, or well-understood changes that benefit from structured discovery and issue decomposition.

Quickly create a single issue from conversation context. Use when you want to create a bead, file an issue, track this as a bead, make a ticket for something, or capture a work item. Lightweight alternative to issue-plan for one piece of work without full planning.

Reconcile local git state after a PR has merged upstream. Use when the user says "the PR merged", "PR is merged, clean up", "pull down main", "reconcile after merge", "clean up the worktree", "wrap up this branch", or any variation of "post-merge cleanup". Use this skill proactively whenever a user mentions that a PR from an /implement worktree has landed and local state needs to catch up - even if they only ask for part of it (e.g. just "pull main"). This skill exits any active worktree, removes it, deletes the merged feature branch, fast-forwards the default branch from the remote, and sweeps /implement artifact directories from /tmp.

Watch a pull request and react locally to upstream events - CI failures, reviews, approvals, merge conflicts, and the merge itself - without ever writing back to GitHub. Use when the user says "watch this PR", "keep an eye on PR #N", "monitor my PR", "let me know when CI passes", "tell me when someone reviews", "alert me when it merges", or any variation of wanting a background observer on a PR. Use this skill proactively whenever a user has just pushed a PR and wants hands-off follow-through while they work on something else. This skill uses the Monitor tool to poll gh on an interval and fires LOCAL-ONLY reactions (investigate failures, import reviews, start rebases, surface notifications). It NEVER posts comments, creates reviews, merges PRs, or pushes commits.

Plan issues by interviewing the user first, then validating with Codex technical review. Use when the user wants to be involved in shaping requirements but also wants AI-powered gap analysis before creating beads. Triggers on "plan with me", "let's think through this", "plan and validate", or "plan with codex review".

Plan issues by interviewing the user with probing questions - no AI review, purely human-driven planning. Use when you want to work together to create beads, be interviewed about requirements, or plan without codex. Best when the user has strong opinions, wants full control, or when domain knowledge lives with the user rather than in the codebase. Use for subjective decisions, product direction, or when Codex review would add noise rather than value.

Post review findings as PR comments on specific lines. Accepts a review report (from /team-branch-review, /team-commit-review, or pasted), interviews the user on which findings to comment on, then posts targeted review comments via GitHub API.

Generate weekly AMPS Bites updates for the Applied Training team by pulling GitHub and Slack activity. Use when the user mentions AMPS bites, weekly update, team status update, weekly summary, or wants to generate an AMPS bites draft. Also trigger when the user says things like "what did we do this week" or "summarize the team's work".

Scan Slack, Twitter/X, and Claude Code changelogs for new skills, features, patterns, and config ideas - then compare against current setup and suggest changes. Also detects staleness in existing rules and skills. Use when the user says things like "what's new in claude code", "scan for CC updates", "check for config improvements", "any new skills this week", "is my setup stale", or "cc scout". Run this periodically (weekly or biweekly) to stay current.

Reply to PR comments after fixes are applied via team-branch-fix

Import PR review comments into a review report for team-branch-fix

Review session work and optionally update agent memory. Use to capture decisions, patterns, or insights - or just to reflect without changing anything.

Set up CoreWeave development environment and manage local CKS clusters. Use when installing development tools, configuring authentication, or starting/stopping local Kubernetes clusters.

Discover what the cw CLI can do for you. Use when learning about cw CLI capabilities, exploring available components and archetypes, or deciding what features to add to a repository.

Guide through creating a new CoreWeave GitHub repository with proper setup. Use when creating new repos with standardized templates, team permissions, and Backstage integration.

Add components to existing repos or generate archetypes locally. Use when adding GitHub workflows, CODEOWNERS, Helm charts, Backstage catalog files, or generating new project structures.

Reimplement the current branch on a new branch with clean, narrative-quality git commit history suitable for reviewer comprehension. Use when you want to clean up messy commit history before opening a PR.

Remove signs of AI-generated writing from text. Use when editing or reviewing text to make it sound more natural and human-written. Based on Wikipedia's comprehensive "Signs of AI writing" guide. Detects and fixes patterns including: inflated symbolism, promotional language, superficial -ing analyses, vague attributions, em dash overuse, rule of three, AI vocabulary words, negative parallelisms, and excessive conjunctive phrases.

Clone and explore external GitHub repositories to understand how libraries, frameworks, or dependencies work. Use when user provides a GitHub URL (github.com/owner/repo), asks "how does X library work", wants to look at source code for a dependency, asks about implementation details of an external package, or says "explore", "look at", or "check out" a repository. Automatically checks out the matching version tag when the repo is a dependency in the current project.

Write comprehensive, user-focused documentation following the Diataxis framework. Use this skill when creating or improving tutorials, how-to guides, reference documentation, or explanatory content. Helps identify the right documentation type and apply best practices for each.

Comprehensive Kubernetes cluster management skill. Use this skill when working with Kubernetes resources, kubectl operations, Helm charts, container orchestration, debugging pods, managing deployments, or any Kubernetes-related infrastructure tasks.