
Loads a written implementation plan, reviews it critically, executes all tasks sequentially, and reports when complete — without improvising beyond the plan's scope.

Skill rank progression over time. Hover for details.
All named implementations attributed to @obra in the registry.

Loads a written implementation plan, reviews it critically, executes all tasks sequentially, and reports when complete — without improvising beyond the plan's scope.

Guides completion of development work by verifying tests first, then presenting structured options for merge, PR creation, or cleanup — never declaring done without a passing build.

Use when implementing any feature or bugfix, before writing implementation code

Finds the root cause before attempting any fix — building a minimal reproduction, forming ranked hypotheses, and instrumenting to confirm before writing a single corrective line.

Use when creating new skills, editing existing skills, or verifying skills work before deployment

Explores user intent, requirements, and design before implementation. Turns ideas into fully formed designs and specs through natural collaborative dialogue.

Executes plans by dispatching a fresh subagent per task with a two-stage review (spec compliance then code quality), keeping the orchestrator context clean throughout.

Writes comprehensive implementation plans that break multi-step tasks into bite-sized, independently executable sub-tasks before any code is touched.

Requires running verification commands and confirming their output before claiming any work is complete, fixed, or passing — no claim without evidence.

Dispatches a code reviewer subagent with isolated context to catch issues before they cascade — producing a structured review report that guides targeted fixes.

Delegates independent tasks to specialized agents with isolated context to work concurrently, dramatically reducing wall-clock time on multi-part problems.

Processes code review feedback with technical rigor and verification before implementation — questioning unclear feedback and confirming fixes with tests rather than blindly applying suggestions.

A complete software development methodology for coding agents — brainstorming, planning, parallel execution, systematic debugging, code review loops, and branch discipline unified into a single agentic workflow.

Ensures every piece of feature work begins in an isolated workspace — creating a native git worktree when available, falling back to tool-provided isolation otherwise.
Want to add more skills?
Register your repo →