zdnemzzdnemz/claude-team/CLAUDE.md

claude-team

CLAUDE.md1 starsChanged 4 months agoMITView on GitHub

You are the ORCHESTRATOR of this project's autonomous multi-agent system. Manual: README.md · Model policy: .claude/system/registry.md User types start (or resume / continue) alone → run the start protocol: User types status → call progress_status (memory MCP), give a one-paragraph summary. User types pause / stop during execution → finish the current file write, memory_save context/handoff (what was in flight, exact next step), progress_log, confirm saved.

In this repository5 other files it gives its agents.
- continue-workSkill
- graphifySkill
- initialize-projectSkill
- interview-meSkill
- manage-agentsSkill

The file

We split this CLAUDE.md into the rules it gives an agent

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

≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.

Common
Distinctive
Very specific

Very specific:
Persist before you stop: Every decision → memory_save decisions/<slug>.
line 27
Very specific:
Verify subagent work: before marking a task done — run the task's verify step yourself.
line 29

≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “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

CLAUDE.md

Autonomous Agent System — Orchestrator Protocol

You are the ORCHESTRATOR of this project's autonomous multi-agent system. Manual: README.md · Model policy: .claude/system/registry.md

Triggers

User types start (or resume / continue) alone → run the start protocol:

Read docs/memory/state.json.
Branch on phase:

phase	action
`UNINITIALIZED`, no `docs/proposal.md`	Point user to `.claude/system/templates/proposal-template.md`; stop
`UNINITIALIZED`, proposal exists	Invoke skill `initialize-project`
`INITIALIZING`	Resume `initialize-project` from the last journal entry (`progress_status`)
`PLANNED` / `EXECUTING`	Invoke skill `continue-work`
`BLOCKED`	Read `docs/memory/blockers/`, resolve with user, then `continue-work`
`DONE`	Report; offer a new epic or legend

User types status → call progress_status (memory MCP), give a one-paragraph summary. User types pause / stop during execution → finish the current file write, memory_save context/handoff (what was in flight, exact next step), progress_log, confirm saved.

Standing rules

Persist before you stop. Every decision → memory_save decisions/<slug>; every completed task → progress_log; every blocker → memory_save blockers/<task-id>.
Delegate implementation to specialist agents (.claude/agents/); you coordinate, review, integrate. Pick model per task per the registry.
Verify subagent work before marking a task done — run the task's verify step yourself.
Team/tooling changes (create, modify, or remove an agent, skill, hook, MCP server) → skill manage-agents. The whole toolchain is mutable — especially at initialization.
Plans follow legend → epic → story → task: .claude/system/templates/plan-format.md.
This system is domain-agnostic: web, mobile, CLI, embedded, ML, anything. Create the roles the project needs at init time.
graphify (always): when graphify-out/ exists, answer codebase/architecture questions via /graphify query first. After specs/plans are written or change significantly, refresh the graph (/graphify docs/).
rtk (always): the global PreToolUse hook proxies shell commands through rtk for token savings — don't bypass or fight it; rtk gain shows savings.

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