Tool-agnostic 13-phase AI development pipeline — turns a task description into reviewed, committed code through automated design, adversarial review, security, test, and code-review gates. One bash engine, balanced Opus/Sonnet routing, self-healing commit review.
Agent Claude Code
These agents back the per-phase slash commands (/design, /ar, /security-review, …) when you run them interactively in Claude Code. The pipeline engine (run-pipeline.sh) does not use them — it inlines every phase prompt in buildprompt(). Editing an agent changes the interactive helper only, never engine behavior.
Agent Claude Code
Multi-perspective critique to stress-test designs. Runs 3 critic passes with different viewpoints to surface blind spots.
Agent Claude Code
Create first-principles design grounded in live documentation research. Every decision must cite authoritative sources.
Agent Claude Code
Create deterministic implementation specs with zero ambiguity. Builders should never need to make decisions.
Agent Claude Code
Execute implementation plans with context isolation per step. No improvisation - follow the plan exactly.
Agent Claude Code
Lightweight proactive code analysis — scan for issues, missing tests, and improvement opportunities without running the full pipeline.
Agent Claude Code
Remove debug artifacts, console.logs, commented code, and development leftovers from production code.
Agent Claude Code
Verify plan-vs-design alignment before build. Catch missing coverage, scope creep, and contradictions.
Agent Claude Code
Reviews implementation plans for completeness, feasibility, risks, and alignment with project conventions. Provides a verdict and actionable feedback.
Agent Claude Code
Creates detailed implementation plans for features, bug fixes, and refactors. Explores the codebase, identifies files to change, and produces a step-by-step plan with acceptance criteria.
Agent Claude Code
Research existing solutions before building anything new.
Agent Claude Code
Behavior validation - verify tests pass and outputs match design specifications.
Agent Claude Code
Documentation coverage check - verify Swagger docs, JSDoc comments, and API documentation.
Agent Claude Code
Code quality check for types, lint, and project conventions. Verify code fits RDO patterns.
Agent Claude Code
Extract and crystallize requirements through structured Q&A. Transforms fuzzy requests into precise, actionable briefs.
Agent Claude Code
Security audit for OWASP vulnerabilities, auth bypass, secrets exposure, and injection attacks.
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: