Assemble the Surgical Team

When invoked, do the following steps in order.

Step 1: Orient

Invoke the using-brooks-team skill to establish team framework context.

Step 2: Survey the Project

Read the current project state:

• Check for AGENTS.md, COPILOT.md, CLAUDE.md, README.md, or any project documentation
• Look at the top-level directory structure
• Check git log --oneline -10 if this is a git repository
• Identify the primary language and framework in use

If there is no project context (empty directory), skip to Step 3 and note that no project was found.

Step 3: Present the Team Briefing

Present a project-contextual briefing in this format:

Your Surgical Team for [Project Name or "New Project"]

Your role: Surgeon (Chief Programmer) You own all critical design and implementation decisions. The team supports you.

Copilot — Code review and backup

• Invoke: copilot skill (inline review), or dispatch the copilot custom agent (subagent review)
• Best for: [describe the main feature areas or recent work in this project where review would help]
• Trigger: After completing any feature of non-trivial size; before any merge

Tester — Adversarial test strategy

• Invoke: tester skill, or dispatch the tester custom agent
• Best for: [describe any apparent testing gaps or recently implemented features]
• Trigger: Before calling any feature complete; when a bug escaped testing

Administrator — Task and project tracking

• Invoke: administrator skill; in Codex, dispatch the administrator custom agent for parallel coordination
• Best for: [describe current project complexity — number of open tasks, known blockers]
• Trigger: Multi-task sessions; unclear priorities; scope creep

Editor — Documentation and clarity

• Invoke: editor skill
• Best for: [describe current documentation state — README present/absent, API docs, etc.]
• Trigger: Before releases; when docs lag implementation; commit message review

Program Clerk — Code organization

• Invoke: program-clerk skill; in Codex, dispatch the program-clerk custom agent for parallel structure review
• Best for: [describe any structural concerns observed — large files, unclear naming, etc.]
• Trigger: Files growing too large; inconsistent naming; tangled imports

Toolsmith — Custom tool builder

• Invoke: toolsmith skill
• Best for: [describe any repetitive operations observed in the project]
• Trigger: Same manual operation performed 3+ times; workflows that take >5 minutes

Language Lawyer — Language and framework expertise

• Invoke: language-lawyer skill (inline), or dispatch the language-lawyer custom agent (subagent investigation)
• Best for: [describe language/framework in use and any version-sensitive areas]
• Trigger: Edge cases; version differences; subtle API behavior; security-sensitive operations

Step 4: Prompt for Next Action

End with:

Which role would you like to activate first, or shall we proceed directly to the work?

If you have a task in mind, describe it and I'll identify which team members should be involved.

Want parallel support?

• In Codex: explicitly ask me to spawn the relevant custom agents, then use /agent to inspect or steer their threads.
• In Grok Build: use the assemble-with-grok-team skill for native parallel subagents.
• In Copilot CLI or OpenCode: use the assemble-with-fleet skill to spawn independent sessions per role.
• In Claude Code: use /assemble-with-agent-teams for Agent Teams.
• In Hermes Agent: use the assemble-with-hermes-team skill for delegate_task-based parallel spawn, tracked on the native kanban board if your profile has the kanban toolset enabled (otherwise a plain plan, same as Codex/OpenCode).
• In Pi Coding Agent: use the assemble-with-pi-team skill if the optional subagent example extension is installed (real parallel dispatch, OS-process based); otherwise pi-coding-agent has no built-in parallel-agent mechanism, so stay single-session here.