Expert in AI/ML pipelines, LLM integration, RAG systems, embeddings, and intelligent automation. Use for building AI-powered features, prompt engineering, and model integration. Triggers on ai, ml, machine learning, llm, gpt, embeddings, rag, vector, chatbot, agent.
Deep, sourced web-research engine. Confirms every datum with ≥2 independent sources, extracts verbatim quotes, flags contradictions and unknowns. Output: findings+provenance JSON artifact (written to file) + one-line return. Shared worker for ds-research and ds-brief.
Expert code quality reviewer for newly created/modified code. Proactively checks correctness, security, maintainability, and test coverage. Use proactively after code changes.
Coding agent via Codex CLI. Use after planning to delegate implementation tasks — feature building, bug fixes, refactoring. Gathers context, formulates a targeted Codex prompt, and runs the implementation.
Confidence-based code review specialist. Use when reviewing code changes, pull requests, or verifying quality before merge. Applies scoring threshold of 80+ to avoid noise.
Analyzes DevExperience history every 10 tasks. Identifies specific tasks needing correction, passes explicit task IDs to FixPlanner. Updates developer-addendum only (never overwrites developer.md).
Delegate to this subagent when you need changeset@2 artifacts produced from a git diff. Input is a git diff, optionally a linked spec@1, and optionally smith's criteriaevidence from the TDD run that produced this diff. A diff spanning multiple independent topics produces one changeset@2 per topic, never one bundled…
Use this agent when you want to eliminate code duplication across your codebase. Examples: Context: User has just written several similar functions and wants to clean up duplication. user: 'I just added these three validation functions that seem really similar' assistant: 'Let me use the code-deduplicator agent to…
Expert code reviewer. Proactively reviews code changes for quality, security, and best practices. Use after implementing features or fixing bugs to ensure code quality.
Validates task files against task template and task-creator rules. Reads sources of truth, checks structure, content quality, and consistency. Triggers: after task-creator generates files, on re-validation after fixes. Not for: security (security-auditor), spec coverage (completeness-validator).
A decision-making helper for an agent orchestrator, used at important points such as architecture choices, difficult bug investigations, and plan reviews.
Build and TypeScript error resolution specialist. Use PROACTIVELY when build fails or type errors occur. Fixes build/type errors only with minimal diffs, no architectural edits. Focuses on getting the build green quickly.
Standard implementation agent for well-specified tasks tagged [T1] — features from a clear spec, unit tests, documentation, renames, config changes, mechanical multi-file edits, straightforward bugfixes with a known repro. MUST BE USED for [T1] tasks.
Agentic design pattern architect. Recommends the optimal combination of patterns from the 28-pattern library for a given problem. Use when: designing a new AI agent system, choosing patterns, comparing pattern trade-offs, planning multi-pattern architectures.
AI agent manager — the host/main agent in any agentic CLI tool. Takes the client's request, routes it into the dev-team roster, runs the collaboration protocol (perspectives → sharing → master plan → task DAG → execution → validation), never implements the deliverable itself.
TEMPLATE for a project-specific implementation agent. Copy this file, rename it, and fill in the placeholders to define a dispatchable role for ONE project: the expertise it brings and how it works. The project's own stack, structure, and commands stay in that repository's root AGENTS.md, which this agent reads rather…
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: