agentic-tools

Build a throwaway prototype to answer a design question. Use when the user wants to sanity-check whether a state model or logic feels right, or explore what a UI should look like.

Splits a large branch or PR into smaller, independently reviewable units by grouping changes and proposing parallel PRs or a stacked series. Use when a changeset is too big to review well and should be broken up.

Captures the portable, non-secret pi config from a live ~/.pi/agent into this repo's pi/ directory and commits it, without reading or copying secrets. Use when live pi config or extensions have drifted from the repo and need to be synced back.

Refreshes PR titles and descriptions to match the current branch changeset. Use when the PR title or body has drifted from the code.

Reviews the current session for problems and their eventual fixes, then proposes durable improvements to repo skills, scripts, or docs that prevent recurrence. Use when a session hit avoidable friction worth turning into a lasting fix.

Runs discovery and uncertainty resolution for a change, then writes a terse plan document covering purpose, behaviour, validation, architecture, and phased vertical slices. Use before building when a change needs a clear, agreed design.

Turns free-form reports into ordered, execution-ready implementation checklists. Use when analysis must become an actionable plan.

Runs the full PR checklist by chaining the pr-* skills in a logical order. Use when a branch should be taken end-to-end to a healthy, review-ready PR.

Implements scoped changes from context, plans, or speclists through verifiable phases with up-front success criteria and per-phase validation. Use when execution should start from existing requirements.

Creates a new git branch for current or proposed changes, deriving a conventional name when none is given. Use when the user wants to start a branch, e.g. `/branch <name?>`.

Creates new repo-native agent profiles from plan to scaffolded `agents/<name>.md` and catalog alignment. Use when adding an agent profile to this repository.

Creates new repo-native skills from plan to scaffolded `SKILL.md` and catalog entry. Use when adding a skill to this repository.

Keeps responses concise and outcome-first by removing redundancy and non-essential detail. Use when writing docs, skills, comments, and chat responses.

Creates a PR for the current branch or worktree when none exists. Use when a branch needs a pull request opened.

Creates or reuses a git worktree for a branch and reports its ready-to-use path. Use when the user wants to work on a branch in its own worktree, e.g. `/worktree <branch-name>`.

Audits a given scope (skill, system prompt, tool set, CLAUDE.md, or project) against context engineering best practices and applies improvements. Evaluates signal-to-noise ratio, system prompt altitude, tool design, retrieval strategy, and long-horizon technique selection.

Checks proposed skill names for collisions with reserved or existing names. Use when validating a skill or command name before creation.

Runs an aggressively strict maintainability review focused on abstraction and complexity risk. Use when a harsh deep quality audit is requested.

Finds and ranks meaningful duplication and DRY violations in a chosen scope. Use when deciding what repeated logic should be consolidated.

Recommends best-fit refactoring techniques for smell findings with confidence. Use when smell findings need scoped refactoring direction.

Runs a prioritized principles review for plans and source artifacts using evidence-based pass/warn/fail outcomes.

Orchestrates smell-to-plan flow by validating findings and assembling a final plan. Use when coordinating multi-smell refactor planning.

Runs structured code-smell analysis and maps findings to standard smell categories. Use when prioritizing maintainability risks and refactors.

Evaluates Single Responsibility Principle violations in a chosen scope and recommends cohesive boundaries. Use when reviewing mixed responsibilities in files, classes, or modules.

Inspects staged and unstaged changes and groups them into coherent change sets. Use when reviewing what changed before committing.

Resolves in-progress merge/rebase/cherry-pick/revert conflicts and continues the git operation. Use when a git operation stops on conflicts.

Rebases one branch onto a new base with conflict delegation and optional force-push lease. Use when a branch needs safe direct rebasing.

Investigates and fixes failed PR CI jobs end-to-end for the current branch. Use when branch CI failures need resolution.

Resolves unresolved PR review comments with code changes, validation, and thread follow-up. Use when review feedback must be addressed end-to-end.

Finds the PR for the current branch (or URL), verifies it, and returns key metadata. Use when PR context is needed before PR work.

Remove AI writing patterns from prose. Use this skill when writing, drafting, editing, reviewing, or revising any text to eliminate predictable AI tells, slop, and formulaic patterns.

Identifies extraction opportunities from DRY and SRP findings and proposes reusable boundaries. Use when planning concrete code extractions in a chosen scope.

Confirms and normalizes analysis scope, then returns clear scope assumptions. Use when work needs explicit boundaries before scanning files.

Creates well-formed commit(s) from current changes with grouped diffs and clear messages. Use when current work is ready to commit.

Creates GitHub issues with normalized fields, optional metadata, and verification. Use when opening an issue, including optional parent linking.

Rebases the current branch onto its PR base and force-pushes with lease. Use when syncing a PR branch with its target base.

Restacks dependent branches after upstream rebase/merge/force-push changes. Use when a PR stack needs bases realigned.