Millsondylan/Claude-code-agents

Millsondylan/Claude-code-agents

Agent Claude Code

Integration testing specialist that verifies components work together correctly. Runs integration tests, checks API contracts, and validates end-to-end workflows.

Millsondylan/Claude-code-agents

Agent Claude Code

Verifies code logic correctness using deep analysis. Detects algorithmic errors, off-by-one bugs, race conditions, edge cases, and logical flaws. Read-only verification.

Millsondylan/Claude-code-agents

Agent Claude Code

Stage 1 meta-orchestrator. Analyzes raw user request complexity to determine how many full sequential pipeline runs are needed. Runs BEFORE prompt-optimizer.

Millsondylan/Claude-code-agents

Agent Claude Code

Creates batched implementation plan with feature assignments. Use after code-discovery to plan implementation.

Millsondylan/Claude-code-agents

Agent Claude Code

Pre-implementation sanity checks before build-agent starts. Verifies environment, dependencies, and prerequisites are ready. Fast validation to catch issues early.

Millsondylan/Claude-code-agents

Agent Claude Code

Updates project-specific sections in CLAUDE.md and ACM. Can ONLY modify PROJECT-SPECIFIC sections (between markers), NEVER base rules.

Millsondylan/Claude-code-agents

Agent Claude Code

Intercepts and optimizes all prompts before they reach target sub-agents. Runs first, outputs only the optimized prompt.

Millsondylan/Claude-code-agents

Agent Claude Code

MANDATORY. Reviews changes against acceptance criteria. Checks for anti-destruction violations (overwrites, unnecessary files, placeholder tests). Read-only.

Millsondylan/Claude-code-agents

Agent Claude Code

ALWAYS FIRST. Analyzes user requests and creates structured TaskSpec with features, acceptance criteria, risks, and assumptions. Use this agent to start any task pipeline.

Millsondylan/Claude-code-agents

Agent Claude Code

MANDATORY. Runs test suite and reports results. NEVER blocks pipeline - always requests debugger on failure. Detects placeholder tests.

Millsondylan/Claude-code-agents

Agent Claude Code

Writes comprehensive, real, fully functional tests with 100% coverage for implemented features. NO mocks, NO placeholders, NO assert True, NO pass stubs. Maps every test to TaskSpec acceptance criteria.

Millsondylan/Claude-code-agents

Agent Claude Code

DEPRECATED - use docs-researcher instead. Researches APIs, frameworks, and syntax patterns via web search.