dream-team

Execute an implementation plan — reads a spec file, detects mode from frontmatter, runs the appropriate strategy

Systematic debugging — reproduce the issue, investigate root cause, apply targeted fix, verify resolution

Start here — brainstorm and plan a new feature or task through interactive conversation

Write a delegated execution spec from the current brainstorming session

Write a sequential execution spec from the current brainstorming session

Write a team execution spec from the current brainstorming session

Use this agent for design decisions, architectural analysis, solution proposals, and structural recommendations. Read-only — produces designs, not code.

Use this agent when code needs to be written, files created, or features implemented. Executes ONE task at a time with full read/write access. Uses TDD. Follows specs exactly.

Use this agent for systematic debugging. Reproduces issues, forms hypotheses, investigates root causes, applies targeted fixes, and verifies resolution. Interactive and methodical — understand it before fixing it.

Use this agent for codebase exploration, documentation research, pattern investigation, and gathering context. Read-only — cannot write or edit files.

Use this agent for qualitative code review against specs and coding standards. Categorizes issues by severity. Read-only — cannot modify files.

Use this agent for proactive security auditing of code changes. Works through a structured 7-category checklist on every review. Read-only — cannot modify files.

Use this agent for independent adversarial testing, integration tests spanning multiple components, and security/boundary testing. Writes test files only — cannot modify source code. Complements builder TDD by testing from the spec (not the implementation) and targeting edge cases builders miss.

Use this agent for final mechanical verification. Runs validation commands, checks acceptance criteria, produces pass/fail report. Read-only.

Use when the user wants to execute an implementation plan. Reads a spec file, detects the execution mode (sequential, delegated, or team) from frontmatter, and runs the appropriate strategy. Pass the spec file path as an argument.

Use when encountering a bug, test failure, unexpected behavior, or when the user asks for help debugging. Hypothesis-driven — reproduce, investigate, fix, verify. Standalone — works outside the plan/spec/build workflow.

Use when the user wants to build something new or plan a feature. Collaborative brainstorming that explores requirements, constraints, and approach through natural dialogue before any spec is written.

Use after /dream-team:plan to write a delegated execution spec (orchestrator + sub-agents). No arguments needed — picks up conversation context from the brainstorming session.

Use after /dream-team:plan to write a sequential execution spec. No arguments needed — picks up conversation context from the brainstorming session.

Use after /dream-team:plan to write a team execution spec (parallel Claude instances). No arguments needed — picks up conversation context from the brainstorming session.

3 hooks across 3 events