Implement features, fix errors, optimize against metrics, bootstrap projects, work test-first, and manage git and worktrees.

Build & execute

These skills do the work: implement plans, fix what's broken, optimize against a metric, and manage the git surface around your changes.

Command	What it does
`/wield:feature-building`	Implement features, plans, and fixes with a structured workflow (`--tdd` for tests-first).
`/wield:release-pipeline`	Merge main → test → review → commit → push → PR in one command.
`/wield:error-fixing`	Fix bugs, test failures, and CI/CD issues with intelligent routing.
`/wield:optimization-loop`	Autonomously iterate against a metric, auto-keep/discard changes.
`/wield:project-bootstrap`	Bootstrap a new project: research, stack, design, plan, implement.
`/wield:test-driven-development`	Build new behavior test-first with a red-green-refactor workflow.
`/wield:git-workflow`	Git operations with conventional commits and auto-split.
`/wield:worktree-management`	Create, inspect, and clean isolated git worktrees.

Implement features, plans, and fixes through a structured pipeline. Use for feature development, plan execution, and code implementation. This is the main "build it" skill — point it at a plan from /wield:implementation-planning. Add --tdd to build tests-first per phase — write tests for current behavior before refactoring, then verify they still pass after implementation.

`/wield:release-pipeline`

The full ship pipeline in one command: merge main, test, review, commit, push, open PR. Use for shipping official releases to main/master or beta releases to dev/beta branches — from a feature branch to a PR URL.

`/wield:error-fixing`

Fix bugs, errors, test failures, and CI/CD issues with intelligent routing. Use for type errors, lint issues, log errors, UI bugs, and general code problems — it routes to the right diagnostic path automatically.

`/wield:optimization-loop`

Run an autonomous iterative optimization loop: N iterations against a mechanical metric, learning from git history, auto-keeping or discarding each change. Use for improving measurable metrics — coverage, performance, bundle size — through repeated experimentation.

`/wield:project-bootstrap`

Bootstrap a new project end-to-end: research, tech stack, design, planning, and implementation. Modes: full (default interactive), auto (autonomous), fast (skip research), and parallel (multi-agent).

`/wield:test-driven-development`

Build new behavior test-first with red-green-refactor: write one failing test, make the smallest change that passes it, then refactor while green. Use test-generation instead when you need to run or assess an existing suite.

`/wield:git-workflow`

Git operations with conventional commits — staging, committing, pushing, PRs, merges. Auto-splits commits by type and scope and scans for secrets before they land.

`/wield:worktree-management`

Create, inspect, and clean isolated git worktrees. Use for feature isolation, worktree health audits, stale cleanup, and monorepo or submodule workflows.

In Claude Code, /wield:feature-building can pair with the wield:fullstack-developer agent. Codex skills use native subagents when they delegate work.