Claude Skill MartBrowse skillsQuick linesLearn by videoTerminal guideWhat is a Skill?
 
← Back to list

Agent Teams Command

Admits a multi-agent team only when parallelism truly pays, then runs it with exclusive ownership, typed IPC and serial evidence-gated integration.

AutomationAdvanced★ 136⑂ 19AI score 8/10Last updated: Aug 9, 2026

What it does

  • Team admission check: computes the orchestration tax (setup, context duplication, IPC, merge conflict, review, cleanup) and keeps a single process when one process can finish inside the same review budget.
  • Command program: objective, non-goals, checkpoint cadence, interrupt policy, state/artifact paths, IPC schema, allowed/denied tools, verifier, stop and recovery conditions, plus one versioned shared-intent surface owned by the commander.
  • Territory isolation: one owner per task, separate worktrees/branches or disjoint files, commander-owned shared schemas until published.
  • Typed IPC: only structured messages (task_id, state, artifact, artifact_hash, evidence, decision, unknowns, termination_reason, next_action) change state — status chatter does not.
  • Serial integration and gates: integrate one verified stream at a time in dependency order, with independent red-team review, human approval before production/spend/destructive actions, rollback prepared first, and cleanup receipts.
  • Failure protocol: NEEDS_INPUT, BLOCKED_DEPENDENCY, BLOCKED_PERMISSION, VERIFY_FAILED, NO_PROGRESS, BUDGET_STOP, with max_attempts: 2.

Who it's for

  • Developers whose sub-agent fan-outs turned into overlapping writes and merge chaos.
  • Agent-pipeline designers who want genuine builder/evaluator separation instead of self-certification.
  • Tech leads coordinating large refactors or migrations with several independent workstreams.
  • Teams that must document approval and rollback paths before shipping or spending.

Examples

  1. Shared schema file: the commander publishes the contract first, serializes schema edits, and workers consume it read-only.
  2. Right-sizing the team: a six-node dependency graph where only two nodes need agent judgment → recruit two workers; deterministic nodes stay scheduler-owned.
  3. Rubber-stamp critic: if the reviewer just agrees with the builder's rationale, rerun with acceptance criteria, artifact and evidence only.
  4. All green, integration broken: set mission status to VERIFY_FAILED and reject incompatible artifacts instead of averaging worker reports.

· · · Install guide · · ·

Try it now, no install

Paste this into Claude to use the skill without installing anything.

Read the instructions in this file and follow them to help me:
https://raw.githubusercontent.com/Mark393295827/third-brain-v7-skills/HEAD/skills/agent-teams-command/SKILL.md

What I want: (describe your task here)

If Claude can't open the link, open it yourself and paste the contents instead.

↓ If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the skills/agent-teams-command folder from the GitHub repo Mark393295827/third-brain-v7-skills into my ~/.claude/skills/agent-teams-command/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/Mark393295827/third-brain-v7-skills.git /tmp/tb7 && mkdir -p ~/.claude/skills && cp -r /tmp/tb7/skills/agent-teams-command ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Open a terminal.
  2. Clone the repository: git clone https://github.com/Mark393295827/third-brain-v7-skills.git
  3. Create the skills directory if needed: mkdir -p ~/.claude/skills
  4. Copy just this skill: cp -r third-brain-v7-skills/skills/agent-teams-command ~/.claude/skills/
  5. Verify that the referenced files references/ender-palantir-command-patterns.md and references/classic-campaigns.md actually exist in the copied folder; if not, ignore those pointers.
  6. Restart Claude Code and try a prompt such as: "Use agent-teams-command to decide whether this work justifies multiple agents, and draft the command program."
  7. Note: the workflow assumes an environment that can really isolate workers (git worktrees, write permissions) and a human available for approval gates.
View source on GitHub ↗License: MIT