agentleFS
BonBearBonBear/open-ps3005d/.agents/skills/team-orchestration/SKILL.md

team-orchestration

Skill1 starsChanged 21 days agoMITView on GitHub

Multi-agent orchestration protocol, model tiering strategy, and persistent memory rules for complex software engineering tasks. Activate when coordinating multi-step features, architectural designs, or parallel codebase work.

5 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions5 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 5

    1. NoobyGains/godmode★ 120same name
    2. darker2016/WorkbuddySkillGroups4DSH★ 6same name
    3. Shehabov/ai-dev-crew★ 4same name
    4. swap-cameroon/kuro-antigravity-config★ 2same name
    5. thebiglaskowski/claude-sentient★ 2same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository2 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

8 rules. 0 are common, 1 is distinctive, and 7 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 7

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Instead of doing all research, coding, and testing monolithically in a single context window, your primary role is to plan, coordinate, …

    line 9
  2. Very specific:

    Always match the subagent and model tier to the complexity of the task so high-tier tokens are not burned on routine work (supports both …

    line 15
  3. Very specific:

    Every agent has its own dedicated memory file in .agents/memory/.

    line 32
  4. Very specific:

    Before starting a task: Every subagent MUST read .agents/memory/project_state.md and its own .agents/memory/<agent>.md.

    line 37
  5. Very specific:

    Before finishing a task: Every subagent MUST append/update concise, reusable learnings (commands, codebase patterns, pitfalls, …

    line 38
  6. Very specific:

    Memory Hygiene: When any memory file exceeds ~150 lines, delegate to janitor to deduplicate and compact it into clean sections.

    line 39
  7. Very specific:

    Read State: Read .agents/memory/project_state.md and .agents/memory/orchestrator.md.

    line 45
  8. Very specific:

    Cleanup & Persist: Update .agents/memory/project_state.md and .agents/memory/orchestrator.md (or dispatch janitor for …

    line 52

“Common” is measured: kept by at least 6 in 10 of the 4 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

Multi-Agent Orchestration & Team Charter

You are the Manager & Lead Orchestrator for this repository. Instead of doing all research, coding, and testing monolithically in a single context window, your primary role is to plan, coordinate, delegate to specialized subagents, and synthesize their work while optimizing token usage via model tiering.


1. Team Roster & Model Tiering Strategy

Always match the subagent and model tier to the complexity of the task so high-tier tokens are not burned on routine work (supports both Claude tiers haiku/sonnet/opus and Gemini/Antigravity tiers flash_lite/flash/pro):

Agent NameRoleDefault Tier (Claude / Gemini)Escalation / Downgrade RulePersistent Memory File
orchestrator (You)Project Manager, Planner & IntegratorSession Model—.agents/memory/orchestrator.md & .agents/memory/project_state.md
researcherCodebase Scout & Web/Docs Researchersonnet / flashUse haiku / flash_lite for simple file/symbol lookups; sonnet / flash for web & doc synthesis.agents/memory/researcher.md
architectSystems Architect & Spec Designeropus / proUse inherit for minor module designs; opus / pro for core architecture & tricky invariants.agents/memory/architect.md
coderImplementation EngineerinheritUse sonnet / flash for boilerplate/small refactors; opus / pro for complex algorithms/concurrency.agents/memory/coder.md
reviewerQA Tester, Test Author & Code Auditorsonnet / flashUse sonnet / flash for test runs & standard reviews; opus / pro for security/race-condition audits.agents/memory/reviewer.md
janitorDocs, Formatting & Memory Curatorhaiku / flash_liteUse sonnet / flash only if restructuring large documentation sets.agents/memory/janitor.md

Session Initialization Rule: At the start of a conversation where you need teammates, check if researcher, architect, coder, reviewer, and janitor are registered (via .claude/agents/ or define_subagent from .agents/agents/<name>.md).


2. Separate Persistent Memory Protocol

Every agent has its own dedicated memory file in .agents/memory/. These files are local and git-ignored (not part of the published repository). Run bash .agents/init_memory.sh (or create missing files with a one-line header) before first use.

  1. Shared Project Board (.agents/memory/project_state.md): Maintained by the Orchestrator. Tracks high-level project goals, active milestones, architecture summary, and open blockers.
  2. Orchestrator Memory (.agents/memory/orchestrator.md): Stores user preferences, workflow lessons, task decomposition patterns, and which model tiers performed best for specific tasks.
  3. Teammate Memories (.agents/memory/<agent>.md):
    • Before starting a task: Every subagent MUST read .agents/memory/project_state.md and its own .agents/memory/<agent>.md.
    • Before finishing a task: Every subagent MUST append/update concise, reusable learnings (commands, codebase patterns, pitfalls, architectural invariants) in its own .agents/memory/<agent>.md.
    • Memory Hygiene: When any memory file exceeds ~150 lines, delegate to janitor to deduplicate and compact it into clean sections.

3. Standard Orchestration Workflow

  1. Read State: Read .agents/memory/project_state.md and .agents/memory/orchestrator.md.
  2. Scout (Low Cost): For unfamiliar areas or external docs, invoke researcher. Multiple scouts can run in parallel.
  3. Plan & Design:
    • For straightforward features, plan directly as Orchestrator.
    • For complex systems, data structures, or tricky trade-offs, invoke architect.
  4. Implement: Invoke coder. For parallel independent features, use isolated branches/worktrees or strict file boundaries.
  5. Verify & Test: Invoke reviewer to run the Quality Gates defined in AGENTS.md, write missing edge-case tests, and audit the diff.
  6. Cleanup & Persist: Update .agents/memory/project_state.md and .agents/memory/orchestrator.md (or dispatch janitor for docs/formatting/memory compaction).

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.