Skip to main content

File-based multi-agent coordination protocol — per-role queues, layered plugin set for Claude Code, profile-v2 for OpenAI Codex

Project description

greatminds

File-based multi-agent coordination protocol. Per-role queues, atomic mv handoff, layered plugin set for Claude Code and profile-v2 setup for OpenAI Codex CLI.

Status

Alpha0.1.0.dev0. Public foundation commit; CLI entry-points are wired incrementally as scripts are ported from the original /opt/coordination/ layout.

What it is

A fleet-orchestration substrate for agents running in tmux panes:

  • R8 finite-state pipeline — tasks move through typed queues (feature_inbox/, feature_plan/, feature_dev/, feature_ui_dev/, feature_docs/, feature_test/, feature_review/, verified/, …) via atomic mv — no daemon, no broker, no DB.
  • Per-role identity — each agent owns one role (e.g. ARCHITECT-PLANNER, DEVELOPER, TESTER, STAND-KEEPER); roles claim from their queue, hand off to the next.
  • Append-only task files — every transition adds a block; full audit trail.
  • Layered plugins for Claude Code — universal coordination-protocol plugin loads for every role, per-role plugins layer on top.
  • profile-v2 setup for Codex — equivalent per-role config, allowing hot-swapping claude ↔ codex per role from one fleet config.
  • coordd — keystroke pusher daemon that watches per-role wake-files and forwards them to the live tmux pane only when the agent is genuinely idle (heartbeat-freshness guard prevents interrupting active work).
  • Stand evidence + gate-check — tasks marked stand_required cannot reach verified/ until matching stand_done/<id>.yaml records the live-stand result.

Design philosophy

  • Files, not state machines in memory. If the orchestrator crashes, the FS still tells you exactly where every task is.
  • Atomic mv is the only handoff primitive — same filesystem, same volume.
  • Each task is human-readable YAML/Markdown — no opaque blobs.
  • Roles are interchangeable between Claude Code and Codex via the same plugin / skill / data layer.

Install

pip install greatminds   # not yet published — first release pending

License

Apache-2.0. See LICENSE.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

greatminds-0.1.1.tar.gz (165.6 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

greatminds-0.1.1-py3-none-any.whl (227.9 kB view details)

Uploaded Python 3

File details

Details for the file greatminds-0.1.1.tar.gz.

File metadata

  • Download URL: greatminds-0.1.1.tar.gz
  • Upload date:
  • Size: 165.6 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.10.4 {"installer":{"name":"uv","version":"0.10.4","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}

File hashes

Hashes for greatminds-0.1.1.tar.gz
Algorithm Hash digest
SHA256 cfb2052f93f2a03ef4c7edbbb575109a41967da51a00d98aa52f88b31e48e733
MD5 c43c488858e97c27eb7c2f0ef12c5c98
BLAKE2b-256 c5f7183ca5a5fc29daa6209254d71476516892fd6e60d90c6857953ab96ae152

See more details on using hashes here.

File details

Details for the file greatminds-0.1.1-py3-none-any.whl.

File metadata

  • Download URL: greatminds-0.1.1-py3-none-any.whl
  • Upload date:
  • Size: 227.9 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: uv/0.10.4 {"installer":{"name":"uv","version":"0.10.4","subcommand":["publish"]},"python":null,"implementation":{"name":null,"version":null},"distro":{"name":"Ubuntu","version":"24.04","id":"noble","libc":null},"system":{"name":null,"release":null},"cpu":null,"openssl_version":null,"setuptools_version":null,"rustc_version":null,"ci":null}

File hashes

Hashes for greatminds-0.1.1-py3-none-any.whl
Algorithm Hash digest
SHA256 0c1832f9b9e755dcc3e32fe85c61160e3b02266ac393dee3e573ca689aff91e2
MD5 52ce752ed1853fc3834f05a799451057
BLAKE2b-256 e1621d6e816243d63f1da7f26ad6f35471d5ae47b51ed0b54a8a3bc1ff703f2b

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page