agentleFS
Sign inStart free
Cor-IncorporatedCor-Incorporated/claude-code-skills/skills/agent-orchestrator/SKILL.md

agent-orchestrator

Skill5 starsChanged 2 months agoMITView on GitHub

Plan and execute multi-agent team compositions for parallel implementation tasks. Provides team patterns, wave-based execution, cross-review protocols, and TeamCreate vs parallel Agent decision matrix. Use when: planning team composition for complex multi-file features, choosing orchestration strategy for 3+ parallel tasks, setting up wave-based execution with dependencies, coordinating cross-review between agents. Do NOT use for: single-agent tasks, basic parallel execution of 2 independent tasks (rules handle that), Codex CLI delegation, or sequential single-file changes.

12 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

2 rules. 0 are common, 2 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 7 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    No mutations (immutable patterns)

    line 90
  2. Distinctive:

    No security vulnerabilities (OWASP top 10)

    line 92

“Common” is measured: kept by at least 6 in 10 of the 7 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Agent Orchestrator

Plan and execute multi-agent teams. Core rules (team mandatory, compacting resilience, mutual review) are in agents-and-parallel.md (always loaded). This skill provides the HOW: patterns, decisions, workflows.

<important if="planning team composition for a multi-agent task"> ## Step 1: Analyze Plan and Map Teams
Read plan (Plans.md or stated tasks)
  -> Identify independent groups (parallelize)
  -> Identify dependencies (sequence)
  -> Select team pattern from references/team-patterns.md
  -> Map task groups to agent roles
  -> Output Team Plan
</important><important if="choosing between TeamCreate and parallel Agent spawning"> ## Step 2: Choose Creation Method
ConditionTeamCreateParallel Agents
2+ waves with dependenciesYesNo
Mutual review needed between workersYesNo
Independent one-shot tasksNoYes
Long-running (compacting risk)YesRisky
3+ agents needing coordinationYesPossible

TeamCreate Flow

TeamCreate(team_name, description)
  -> TaskCreate (all tasks with descriptions)
  -> Agent (spawn teammates with team_name)
  -> TaskUpdate (assign tasks)
  -> Monitor via messages + TaskList

Parallel Agent Flow

Agent(task-A, run_in_background) +
Agent(task-B, run_in_background) +
Agent(task-C, run_in_background)
  [all in single message]
  -> Wait for completions
  -> Agent(code-reviewer, review all changes)
</important><important if="setting up wave-based execution or monitoring agent progress"> ## Step 3: Execute and Monitor

Track status at milestones:

| Agent | Task | Status |
|-------|------|--------|
| name  | task | status |

Wave Transitions

Wave N complete
  -> Reviewer(s) cross-review all Wave N changes
  -> Fix CRITICAL/HIGH issues
  -> Re-review if fixes applied
  -> Wave N+1 begins
</important><important if="coordinating cross-review between agents after a wave completes"> ## Step 4: Cross-Review Protocol
ScenarioReviewer
Agent A implements XAgent B or dedicated reviewer
Agent B implements YAgent A or dedicated reviewer
Single implementerDedicated code-reviewer required

Review checklist:

  • Code quality (naming, structure, readability)
  • No mutations (immutable patterns)
  • Error handling present
  • No security vulnerabilities (OWASP top 10)
  • Consistent with existing codebase patterns
  • Tests cover the implementation

Escalation: CRITICAL issues -> fix -> re-review -> CRITICAL = 0 to proceed. </important>

<important if="handling agent failures, compacting, or wave execution problems"> ## Error Handling
SituationAction
Agent hangs / no responseCheck TaskList; re-spawn if stuck
Compacting mid-waveSave state to TaskList; resume from checkpoint
Cross-review finds CRITICALBlock wave transition; fix first
Agent produces conflicting changesResolve via dedicated mediator agent
Wave dependency brokenRoll back wave; re-execute with corrected inputs
</important><important if="selecting a team pattern for a new multi-agent task"> ## Team Patterns

See references/team-patterns.md for full details:

  1. Feature Implementation -- Wave-based multi-component features
  2. Bug Fix + Audit -- Fix + codebase-wide pattern search
  3. Refactoring -- Large-scale file modifications
  4. Research + Implement -- Explore then build
  5. Lightweight Parallel -- Independent small tasks with haiku agents</important>

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free