Claude Skills by KjellKod

Play a quest completion celebration animation. Use when the user invokes $celebrate, asks to celebrate a quest, or when a quest reaches the complete/archived state.

Generate commit messages from staged changes by matching repo conventions (Conventional Commits or plain English), leading with intent, and appending the Quest co-author trailer. Use when the user asks for a commit message, help with git commit, or when reviewing staged changes for commit.

Create and update GitHub pull requests in draft mode. Generates PR title and description from all branch commits, shows for approval before executing. Use when the user asks to create a PR, update a PR description, or open a pull request.

Shepherd an existing PR through CI and review feedback, then mark ready for review when clean. Uses inline-first review handling; PR creation belongs to pr-assistant.

Review local staged plus unstaged tracked-file changes before commit. Use when the user invokes $pre-commit-review, asks for a pre-commit review, asks to review local changes before commit, or wants a local working-tree review before a PR exists.

Multi-agent quest orchestration. Plans, reviews, builds, and fixes features through coordinated agent handoffs. Use when the user invokes $quest or asks to run/resume Quest workflow.

Adversarial interview against a plan, design, or write-up. Use when the user invokes $sharpen, says "sharpen this", "stress-test this", "find the holes", "challenge my plan", or wants to confirm shared understanding before locking a decision.

Run the canonical UX stress-test rubric against a target (file, directory, URL, screenshot, or git diff) and produce a structured critique report with P0–P3 severity, principle citations from the bundled UX guidebook, and a bright-spots section. Used by reviewer agents in the quest pipeline when ui_work=true, and invocable directly with /ux-review or $ux-review.

Delegate a task to OpenAI Codex via MCP. Use when the user invokes /gpt, asks to use codex, or wants a second opinion from a different model.

Automated CI code review for GitHub PRs using OpenAI Codex. Validates PR descriptions, checks implementation alignment, enforces Quest architecture boundaries, and maps test coverage to acceptance criteria when PRs move to ready-for-review.

Review actual code implementations (Python, JavaScript, TypeScript, React) for correctness, maintainability, security hygiene, and adherence to project patterns. Use when reviewing pull requests to ensure code quality, proper error handling, adequate test coverage mapped to acceptance criteria, and alignment with architecture boundaries and review standards.

Implement an approved implementation plan step by step, producing small reviewable changes, mapping code/tests to acceptance criteria, and maintaining a lightweight decision log. Use when the plan/spec is already agreed and you want disciplined execution.

Create implementation plans with testable acceptance criteria, validation strategies, integration touchpoints, and risk analysis before coding begins.

Review implementation plans, PR specifications, and feature documentation to ensure acceptance criteria are testable, validation is complete, and integration risks are surfaced before coding begins.

Canonical decision policy for turning review findings into `review_backlog.json` decisions (`fix_now`, `verify_first`, `defer`, `drop`, `needs_human_decision`) with deterministic metadata and loop-cap behavior.

Primer skill that loads the canonical UX guidebook and stress-test rubric, and owns the UX Defaults emission protocol that the planner follows for `ui_work: true` quests. At activation, announce the skill name and scope in one line. Example: `[ux-context] UX guidebook loaded — principles and stress-test rubric in scope.` **Not user-invocable.** Auto-attached by the orchestrator when the router classifies a quest as `ui_work: true`. For direct critique of existing UI, see `.skills/ux-review/SK...

Quest planning sometimes depends on research that is naturally parallel: reviewing external docs, reading local markdown, comparing sibling repos, mapping dependency surfaces, surveying migration costs, or pressure-testing a proposal from multiple perspectives. Today the planner is effectively one research/planning agent. That keeps the pipeline simple, but it underuses parallelism on research-heavy quests. The better direction is a reusable research skill: humans can invoke it directly, and ...