sdlc-1-plan
01Agent Claude Code
Turn the incoming feature request into a concrete implementation plan with explicit, checkable acceptance criteria.
An artificer's kit for building minimal dark-factories on Claude Code. Compose small, gated agent steps into deterministic builds you can read end to end.
Agent Claude Code
Turn the incoming feature request into a concrete implementation plan with explicit, checkable acceptance criteria.
Agent Claude Code
Summarize the whole sdlc run into one concise final report for the chat, drawing on the prior stages' handoffs.
Agent Claude Code
Write Vitest tests for the behavior described by the previous stage's plan, in the RED state where they run and fail because the implementation does not exist yet.
Agent Claude Code
Make the failing tests from the previous stage pass with the minimum correct implementation.
Agent Claude Code
Tidy the implementation from the previous stage without changing its behavior.
Agent Claude Code
Review the changes produced by the earlier stages and report only real, substantiated defects.
Agent Claude Code
Have Codex independently review the changes produced by the earlier stages.
Agent Claude Code
Reconcile the two independent reviews (review-claude and review-codex) and emit a single clear verdict the workflow can branch on.
Agent Claude Code
Exercise the implemented feature in a real browser via the chrome-devtools MCP against the running app at http://localhost:3000 and verify it meets the plan's acceptance criteria.
Agent Claude Code
Commit the changes this workflow produced on the current branch.
Agent
Part of build-workflow
Generator worker for the build-workflow skill. Scaffolds one stage of a multi-step workflow: a per-stage subagent with a SubagentStop gate, a manual launcher skill, and a dynamic-workflow script.
At most 3 mods per repository are shown here, and a mod shipped inside a plugin is left to that plugin's page — the rest are on their repository pages: