Skillget

multi-agents

agent
Guvenlik Denetimi
Basarisiz
Health Gecti
  • License — License: MIT
  • Description — Repository has a description
  • Active repo — Last push 0 days ago
  • Community trust — 84 GitHub stars
Code Basarisiz
  • exec() — Shell command execution in .github/scripts/attribution.test.cjs
  • exec() — Shell command execution in .github/scripts/workflows.test.cjs
  • spawnSync — Synchronous process spawning in .github/scripts/workflows.test.cjs
  • process.env — Environment variable access in .github/scripts/workflows.test.cjs
  • fs module — File system access in .github/scripts/workflows.test.cjs
  • process.env — Environment variable access in .github/workflows/ci-gate.yml
Permissions Gecti
  • Permissions — No dangerous permissions requested

Bu listing icin henuz AI raporu yok.

SUMMARY

Multi-agent teams for real work: run teams of coding agents (Claude Code, Codex, any model) in isolated git worktrees, and measure whether the team beats a solo agent on cost and results.

README.md

multi-agents

Multi-agent teams for real work. This project is becoming a headless runner for teams of coding
agents, benchmarked against solo agents on fresh GitHub-issue tasks (resolve rate, cost, wall
time). The direction, architecture, team-spec grammar and milestones are in
docs/ROADMAP.md; the contracts the runner is built on are in
agent/teams/types.py. The runner and benchmark are still under
construction.

Setup

Python 3.10+ and uv are required.

python3 -m pip install uv
uv sync --locked --extra dev

Workflow patterns

agent/workflows holds the reusable orchestration patterns built
on the OpenAI Agents SDK that the team runner reuses: SequentialWorkflow, RefineLoop,
BestOfN and Router, plus refinement checkpoints and run artifacts. Each takes an injectable
runner, so they can be tested offline. See the workflow guide for
usage.

Development

make test         # ruff + offline Python tests
make check        # also runs the GitHub workflow and attribution tests (Node.js 22+)
make help         # list all targets

None of the checks need API keys.

Archived work

Pre-pivot experiments live in the archived repository
BichengWang/multi-agents-legacy.

License

MIT

Yorumlar (0)

Sonuc bulunamadi