gwt-execute
Standardizes the execution lane for implementation tasks starting from GitHub Issues.
Install
mkdir -p .claude/skills/gwt-execute && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/19110" && unzip -o skill.zip -d .claude/skills/gwt-execute && rm skill.zipInstalls to .claude/skills/gwt-execute
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
Use when execution should start from a GitHub Issue, gwt-spec Issue, or approved standalone task through one build/test/verify loop.Key capabilities
- →Initiate development from GitHub Issues
- →Execute build-test-verify loops
- →Manage design-gated TDD cycles
- →Adopt and recover execution sessions
- →Verify PR readiness
How it works
The skill uses a unified execution lane that maps tasks to GitHub issues or standalone items. It enforces a strict Red-Green-Refactor TDD loop and requires verification evidence before PR creation.
Inputs & outputs
When to use gwt-execute
- →Implementing a feature from a GitHub issue
- →Running a test-driven development loop
- →Advancing a task through build phases
- →Resolving standalone development items
About this skill
When to use
Execute one scoped work item through implementation, verification, and delivery. Detect direct, design-gated, or standalone mode from the owner and its artifacts.
Completion
Keep the same owner through TDD, verify.plan → verify.run, Ready PR, and CI auto-merge. Confirm delivery against the configured base. Preserve execution authority and recovery audits; use No Action only for a proven delivered owner with zero source surface.
With execution.status reporting launch_route: autonomous, resolve reversible,
in-scope choices using the smallest safe default. Do not stop to wait for a PM reply
or say "ready for the next instruction" after posting a decision. Record
PM default ruling (override available): <choice and rationale> on the Board and
continue. After receiving a Board ruling, resume implementation, verification,
PR, or settlement without another pm.message.send. Stop for a decision only at
irreversible decisions or needs_human boundaries. Existing permission and
delivery gates remain in force; temporary admission or CI waits stay active.
References
- When establishing scope, stop conditions, and gwtd resolution: ownership.
- Before lifecycle changes, adoption, repair, or recovery: lifecycle recovery.
- When choosing and executing the owner mode: modes.
- Before verification, acceptance updates, or a PR handoff: verification delivery.
- For host admission, retries, and compatibility aliases: admission.
- For the constraint review: router audit.
When not to use it
- →Spec-implementation mismatches requiring design discussion
- →Environments with missing auth or tooling
- →Tasks requiring design ownership before implementation
Prerequisites
Limitations
- →Requires GWT_BIN resolution before execution
- →Blocked by missing verification evidence
- →Cannot proceed if design surface changes significantly
How it compares
It integrates design-gated mode for spec-tagged issues, ensuring that production edits only occur after design verification, unlike generic task runners.
Compared to similar skills
gwt-execute side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| gwt-execute (this skill) | 0 | 3mo | No flags | Intermediate |
| testing-workflow | 16 | 12mo | Review | Intermediate |
| dev | 2 | 8mo | Review | Advanced |
| overnight-development | 1 | 3mo | Review | Intermediate |
Try saying
Example prompts that trigger this skill in your AI assistant.
More by akiojin
View all by akiojin →You might also like
testing-workflow
amo-tech-ai
Comprehensive testing workflow for E2E, integration, and unit tests. Use when testing applications layer-by-layer, validating user journeys, or running test suites.
dev
atopile
LLM-focused workflow for working in this repo: compile Zig, run the orchestrated test runner, consume test-report.json/html artifacts, and discover/debug ConfigFlags.
overnight-development
jeremylongshore
Automates software development overnight using git hooks to enforce test-driven Use when appropriate context detected. Trigger with relevant phrases based on skill purpose.
resolve-checks
flowglad
Resolve all failing CI checks and address PR review feedback on the current branch's PR. Runs tests locally, fixes failures, incorporates valid review comments, and resolves addressed feedback. Use when CI is red, after receiving PR feedback, or before merging.
sprint
meta-pytorch
Work on a batch of GitHub issues in parallel using Agent Teams. Creates one worktree per issue with TDD enforcement, coordinates via a lead agent, then produces stacked PRs.
testing
hw-native-sys
Testing guide and pre-commit testing strategy for PTO Runtime. Use when running tests, adding tests, or deciding what to test before committing.