What problem does it solve? Manually implementing well-specified GitHub issues is a bottleneck for development teams. This Skill sets up a fully automated pipeline where an AI agent implements labeled issues, opens pull requests, iterates on review feedback, and promotes merged changes — reducing human involvement to writing the issue and one review pass. ## Core Features & Use Cases - Four-Stage GitHub Actions Pipeline: Generates 1-implement.yml, 2-fix-review.yml, 3-merge-develop.yml, and 4-test-promote.yml workflows covering the full Issue → PR → Review → Merge lifecycle. - Agent-Agnostic Configuration: Switch between Claude and Codex agents with a single config line in .github/agent-pipeline.yml, with support for custom target branches, trigger labels, and review iteration caps. - Safety Guardrails: Enforces opt-in labeling, iteration caps, timeouts, PR-only changes, and human escape hatches with a full audit trail. - Use Case: A team with strong CI and a CLAUDE.md wants AI agents to handle routine feature issues. Run the setup operation to generate the config and workflow files, configure secrets, and verify the pipeline with a test issue. ## Quick Start Ask the agent to set up the agent-pipeline workflow in your repository by generating the config file and four GitHub Actions workflow files, then labeling a test issue to verify the pipeline fires.
