Stats
Actions
Tags
From ai-feature-pipeline
A structured multi-agent pipeline for AI-assisted feature development.
How this skill is triggered — by the user, by Claude, or both
Slash command
/ai-feature-pipeline:afp-pipelineThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
A structured multi-agent pipeline for AI-assisted feature development.
prompts/architect.mdprompts/dev.mdprompts/pm.mdprompts/qa.mdprompts/review.mdregistries/analytics-events.mdregistries/paywall-touchpoints.mdregistries/scope-checklist.mdregistries/ship-checklist.mdscripts/agent-runner.tsscripts/rebuild-context.mjsscripts/run-pipeline.shtemplates/dev-log.mdtemplates/feature-brief.mdtemplates/qa-report.mdtemplates/repository-context.mdtemplates/retrospective.mdtemplates/review-report.mdtemplates/technical-plan.mdA structured multi-agent pipeline for AI-assisted feature development. PM → Architect → Dev → Review → QA.
Model-agnostic, stack-agnostic. Works with any AI coding tool (Claude Code, Codex, OpenCode, Cline).
| Command | Description |
|---|---|
new <description> | Start a new feature. Scopes the feature, generates feature-brief.md |
run --role <role> --slug <slug> | Run a single agent role (pm, architect, dev, review, qa) |
pipeline --slug <slug> | Run the full pipeline for a feature, end-to-end |
Each stage produces an artifact in .ai/artifacts/features/<slug>/.
Prompt: prompts/pm.md
Output: feature-brief.md — requirements, acceptance criteria, i18n, analytics, paywall, scope
Prompt: prompts/architect.md
Output: technical-plan.md + repository-context.md — architecture, impacted files, risks, implementation order
Prompt: prompts/dev.md
Output: Code changes + dev-log.md — implements the feature per the tech plan
Prompt: prompts/review.md
Output: review-report.md — checks implementation against the brief
Prompt: prompts/qa.md
Output: qa-report.md — validates E2E flows
Each stage can be executed via the agent-runner CLI:
node scripts/agent-runner.ts --role=<role> --slug=<slug> --project-root=<path>
Or run the full pipeline:
bash scripts/run-pipeline.sh <slug> [--project-root=<path>]
Reference these registries when scoping or reviewing features:
registries/scope-checklist.md — 7 questions (IN/OUT, entry points, edge cases, etc.)registries/ship-checklist.md — pre-MR approval checklistregistries/analytics-events.md — analytics signal registry (project-specific, in .ai/registry/)registries/paywall-touchpoints.md — paywall surface registry (project-specific, in .ai/registry/)The module reads project configuration from .ai/config.json. Run the afp-setup skill to generate it.
Current: 1.0.0
Provides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.
npx claudepluginhub arnaudmanaranche/ai-feature-pipeline --plugin ai-feature-pipeline