{"id":51951748,"url":"https://github.com/dubsopenhub/agent-orchestra","last_synced_at":"2026-07-29T06:01:30.124Z","repository":{"id":356180556,"uuid":"1230265281","full_name":"DUBSOpenHub/agent-orchestra","owner":"DUBSOpenHub","description":"Multi-agent fleet conductor for the GitHub Copilot CLI.","archived":false,"fork":false,"pushed_at":"2026-05-06T23:33:42.000Z","size":11173,"stargazers_count":2,"open_issues_count":0,"forks_count":1,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-07-07T22:21:29.219Z","etag":null,"topics":[],"latest_commit_sha":null,"homepage":"https://dubsopenhub.github.io/agent-orchestra/","language":"Shell","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"mit","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/DUBSOpenHub.png","metadata":{"files":{"readme":"README.md","changelog":"CHANGELOG.md","contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":"SECURITY.md","support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":"AGENTS.md","dco":null,"cla":null}},"created_at":"2026-05-05T20:45:28.000Z","updated_at":"2026-05-21T19:23:53.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/DUBSOpenHub/agent-orchestra","commit_stats":null,"previous_names":["dubsopenhub/agent-orchestra"],"tags_count":0,"template":false,"template_full_name":null,"purl":"pkg:github/DUBSOpenHub/agent-orchestra","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DUBSOpenHub%2Fagent-orchestra","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DUBSOpenHub%2Fagent-orchestra/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DUBSOpenHub%2Fagent-orchestra/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DUBSOpenHub%2Fagent-orchestra/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/DUBSOpenHub","download_url":"https://codeload.github.com/DUBSOpenHub/agent-orchestra/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/DUBSOpenHub%2Fagent-orchestra/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":36020486,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-07-20T02:08:10.276Z","status":"online","status_checked_at":"2026-07-29T02:00:04.910Z","response_time":95,"last_error":null,"robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":true,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":[],"created_at":"2026-07-29T06:01:28.313Z","updated_at":"2026-07-29T06:01:30.095Z","avatar_url":"https://github.com/DUBSOpenHub.png","language":"Shell","funding_links":[],"categories":[],"sub_categories":[],"readme":"**🎼 Agent Orchestra**: Multi-agent fleet conductor for the GitHub Copilot CLI. \n\u003cp align=\"center\"\u003e\n  \u003ca href=\"https://dubsopenhub.github.io/agent-orchestra/\"\u003e\u003cstrong\u003e🌐 Open the live Agent Orchestra site →\u003c/strong\u003e\u003c/a\u003e\n\u003c/p\u003e\n\n**Multi-agent fleet conductor for the GitHub Copilot CLI. \n\nLaunch multiple visible Terminal Stampede commander groups, let them collaborate in real time, score output quality with sealed [Shadow Score](https://github.com/DUBSOpenHub/shadow-score-spec/blob/main/SPEC.md) gates, and judge run repeatability with [Fleet Scorecard](https://github.com/DUBSOpenHub/fleet-scorecard-spec/blob/main/SPEC.md).**\n\n[![GitHub Copilot CLI](https://img.shields.io/badge/platform-Copilot%20CLI-232F3E.svg)](https://docs.github.com/copilot/concepts/agents/about-copilot-cli)\n[![Live Site](https://img.shields.io/badge/live%20site-Agent%20Orchestra-7dd3fc?logo=githubpages\u0026logoColor=white)](https://dubsopenhub.github.io/agent-orchestra/)\n[![License: MIT](https://img.shields.io/badge/License-MIT-green.svg)](LICENSE)\n[![Security Policy](https://img.shields.io/badge/Security-Policy-brightgreen?logo=github)](SECURITY.md)\n\n\u003e ### ⚡ One Command. That's It.\n\u003e\n\u003e **Never used the CLI before? No problem.** Follow these 3 steps:\n\u003e\n\u003e **1. Open your terminal**\n\u003e - 🍎 **Mac:** Press `⌘ + Space`, type **Terminal**, hit Enter\n\u003e - 🪟 **Windows:** Open **Git Bash** or **WSL** (the installer uses `bash`)\n\u003e - 🐧 **Linux:** Press `Ctrl + Alt + T`\n\u003e\n\u003e **2. Paste this line and press Enter:**\n\u003e ```bash\n\u003e curl -fsSL https://raw.githubusercontent.com/DUBSOpenHub/agent-orchestra/main/quickstart.sh | bash\n\u003e ```\n\u003e *Already have the CLI? No worries — this detects what is present and installs the local Agent Orchestra helpers.*\n\u003e\n\u003e **3. Watch the fleet:** `agent-orchestra-pulse`\n\u003e\n\u003e That's it — raise the baton. 🎼\n\u003e\n\u003e *Requires an active [Copilot subscription](https://github.com/features/copilot/plans) for live Copilot CLI fleet runs.*\n\n---\n\n## 🚀 30-Second Overview\n\nAgent Orchestra is for work that is too big, risky, or cross-cutting for one agent thread:\n\n- **Need multiple strategies at once?** It launches commander-led groups in visible terminal panes.\n- **Need live collaboration, not isolated answers?** Commanders propose, review, improve, converge, and broadcast learnings through append-only ledgers.\n- **Need real observability?** Agent Pulse and the Stampede monitor read concise run telemetry as the work happens.\n- **Need quality gates?** [Shadow Score](https://github.com/DUBSOpenHub/shadow-score-spec/blob/main/SPEC.md) criteria are sealed before launch and applied only after commander bundles finish.\n- **Need a run decision?** [Fleet Scorecard](https://github.com/DUBSOpenHub/fleet-scorecard-spec/blob/main/SPEC.md) answers what changed, what won, what failed, and whether to run it again.\n- **Need honest partials?** A partial launch stays partial; Agent Orchestra never silently upgrades incomplete work to success.\n\nNo server. No message broker. No dashboard daemon. Just Copilot CLI skills, Terminal Stampede, tmux, and filesystem IPC.\n\n---\n\n## 📡 Watch Every Run Live with Agent Pulse\n\nAgent Orchestra is best experienced with [**Agent Pulse**](https://github.com/DUBSOpenHub/copilot-cli-agent-pulse), the real-time terminal dashboard for Copilot CLI sessions and swarm runs.\n\n\u003cp align=\"center\"\u003e\n  \u003ca href=\"https://github.com/DUBSOpenHub/copilot-cli-agent-pulse\"\u003e\n    \u003cimg src=\"https://raw.githubusercontent.com/DUBSOpenHub/copilot-cli-agent-pulse/main/assets/dashboard-screenshot.png\" alt=\"Agent Pulse dashboard showing live Copilot CLI sessions, agents, and swarm telemetry\" width=\"850\"\u003e\n  \u003c/a\u003e\n\u003c/p\u003e\n\nAgent Pulse makes Agent Orchestra observable while it runs:\n\n| What you see live | Why it matters |\n|---|---|\n| Commander status | Know which commander groups are running, partial, failed, or complete |\n| Sub-agent counts | Track live, completed, failed, and seen sub-agents without ledger spelunking |\n| Terminal Stampede runs | See active `.stampede/run-*` work across repos |\n| Telemetry confidence | Distinguish authoritative live counts from recent/historical events |\n| Recent launches | Spot bursts, stuck runs, and platform-cap pressure |\n\nInstall it alongside Agent Orchestra:\n\n```bash\ncurl -fsSL https://raw.githubusercontent.com/DUBSOpenHub/copilot-cli-agent-pulse/main/quickstart.sh | bash\nagentpulse\n```\n\n---\n\n## 🐝 Dogfooded at Swarm Scale\n\nAgent Orchestra follows the same swarm-scale operating model that shaped Agent Conductor: visible commander groups, bounded sub-agent fan-out, collaboration ledgers, live dashboard telemetry, Shadow Score sealing, Fleet Scorecard run decisions, install flow checks, and repo-readiness validation.\n\nThose signals come from local Stampede run artifacts and compatibility ledgers, so the system can be tested without adding a server, queue, or hosted control plane.\n\n---\n\n## 🤔 What Is This?\n\nAgent Orchestra coordinates multiple commander-led agent groups against the same mission. Each commander gets its own namespace, bounded sub-agent hierarchy, collaboration bus access, and final bundle contract. The orchestrator then synthesizes the best findings from commander bundles, collaboration ledgers, live telemetry, sealed [Shadow Score](https://github.com/DUBSOpenHub/shadow-score-spec/blob/main/SPEC.md) results, and [Fleet Scorecard](https://github.com/DUBSOpenHub/fleet-scorecard-spec/blob/main/SPEC.md) run decisions.\n\n```text\nYou\n  ↓\nAgent Orchestra\n  ↓\nTerminal Stampede commander panes\n  ↓\nBounded sub-agent groups\n  ↓\nCollaboration bus + live telemetry\n  ↓\nShadow Score + Fleet Scorecard + final synthesis\n```\n\nUse it for final release reviews, architecture audits, migration plans, repo-readiness passes, and high-stakes implementation design where one model answer is not enough.\n\n---\n\n## ⚡ Quick Start\n\n### Install from GitHub\n\n```bash\ncurl -fsSL https://raw.githubusercontent.com/DUBSOpenHub/agent-orchestra/main/quickstart.sh | bash\n```\n\nThe quick installer:\n\n1. Checks for required local tools.\n2. Verifies `python3` and `git` are available.\n3. Installs the Stampede launcher, monitor, Fleet Scorecard, and Agent Orchestra helper launcher to `~/bin/`.\n4. Runs the local smoke gate.\n5. Leaves the repo ready for Agent Pulse observability and fleet validation.\n\n### Install from a local clone\n\n```bash\ngit clone https://github.com/DUBSOpenHub/agent-orchestra.git\ncd agent-orchestra\n./install.sh\n```\n\n### Run\n\n```text\nagent conductor on ~/dev/my-repo : evaluate auth architecture and propose fixes\nagent conductor premium max on ~/dev/my-repo : final release readiness review\nagent conductor standard small on ~/dev/my-repo : review docs and onboarding gaps\nagent-orchestra-pulse\n```\n\nAgent Orchestra uses the Agent Conductor command surface for fleet missions and Agent Pulse for live visibility.\n\n---\n\n## 🧭 Command Reference\n\n| Command | Purpose |\n|---|---|\n| `agent conductor` | Ask only for missing mission/repo; default to Premium Max + Agent Pulse + Stampede monitor |\n| `agent conductor on REPO : MISSION` | Launch with Premium Max + Agent Pulse + Stampede monitor |\n| `agent conductor premium max on REPO : MISSION` | Launch 5 premium commander groups with Agent Pulse + Stampede monitor |\n| `agent conductor standard small on REPO : MISSION` | Launch 5 standard-tier commander groups with Agent Pulse + Stampede monitor |\n| `agent conductor status [RUN_ID]` | Show concise stats, results, and collaboration counts |\n| `agent conductor teardown RUN_ID` | Stop the underlying Stampede tmux session |\n| `agent-orchestra-pulse` | Open Agent Pulse against this repo's fleet telemetry |\n| `bin/fleet-scorecard --repo . --run-id RUN_ID` | Regenerate a Fleet Scorecard for a completed run |\n\n\u003e Note: scale words like `small`, `standard`, and `max` affect tier/policy, not commander count.\n\u003e Agent Orchestra follows the same five-commander fleet contract.\n\n---\n\n## 📊 Live Stats\n\nThe Stampede monitor and [Agent Pulse](https://github.com/DUBSOpenHub/copilot-cli-agent-pulse) read `.stampede/{run_id}/orchestrator-commentary.json` for concise, dashboard-safe stats:\n\n```text\ncmd 3/5 active · sub-agents 112 running / 480 done / 620 seen · q 0 · claimed 3 · results 2/5\ncollab p5 r18 i11 c8 b7\ncommander-004 launching_workers · squads 32/50 · sub-agents 160/250 · run 42 done 118 fail 0\n```\n\nIn chat, Agent Orchestra uses swarm-style commentary: a short phase banner, commander progress table, totals, and a clear decision line (`wait`, `stop`, `synthesize partial`, or `relaunch`). Startup failures are called out as `failed_startup` instead of being described as merely still running.\n\nAgent Orchestra distinguishes:\n\n- **Running now** — live commander, squad lead, and worker telemetry.\n- **Seen** — historical sub-agent launches recorded in the compatibility ledger.\n- **Completed/failed** — terminal outcomes from commander telemetry.\n- **Partial** — incomplete launches or degraded runs that must not be called success.\n\n\u003e 📡 **Best real-time view:** keep **Agent Pulse** open beside **Agent Orchestra** so commander status, sub-agent counts, recent launches, and telemetry confidence stay visible while the run is active.\n\n---\n\n## 👻 Shadow Score\n\nAgent Orchestra creates a sealed [Shadow Score](https://github.com/DUBSOpenHub/shadow-score-spec/blob/main/SPEC.md) envelope before launch. Commanders and sub-agents see only the seal hash, never the hidden criteria. The orchestrator verifies the seal and evaluates final bundles only after the run completes.\n\nShadow Score focuses on:\n\n| Dimension | Question |\n|---|---|\n| Requirement coverage | Did the final synthesis answer the mission? |\n| Collaboration quality | Did commanders review, improve, and adopt useful findings? |\n| Evidence quality | Are claims backed by files, logs, tests, or ledgers? |\n| Validation impact | Did the run improve measurable confidence? |\n| Synthesis usefulness | Did it combine the best work instead of picking vibes? |\n\n---\n\n## 📋 Fleet Scorecard\n\nEvery run also gets a [Fleet Scorecard](https://github.com/DUBSOpenHub/fleet-scorecard-spec/blob/main/SPEC.md)\noverlay. The launcher follows the Fleet Scorecard Spec by sealing a lightweight\nfour-question rubric before commanders start, then teardown verifies the seal and\nwrites `.fleet-scorecards/{run_id}/scorecard.md`.\n\nFleet Scorecard answers:\n\n| Question | Purpose |\n|---|---|\n| What changed? | Captures repo diffs, artifacts, decisions, and useful knowledge |\n| What won? | Names the best commander output, idea, or recommendation |\n| What failed? | Preserves partials, missing bundles, weak evidence, and teardown caveats |\n| Would I run it again? | Gives a clear rerun decision and next-run modification |\n\nThis is intentionally different from cost accounting. It is the final\nhuman-readable judgment layer that turns many commander outputs into one\nrepeatable run decision.\n\nAgent Orchestra is the first **FSS-L4 reference implementation**: run card,\nsealed rubric, evidence index, automatic teardown emission, and source run link.\n\nRegenerate a scorecard manually:\n\n```bash\nbin/fleet-scorecard --repo . --run-id run-YYYYMMDD-HHMMSS\n```\n\n---\n\n## 🏗️ Project Structure\n\n```text\nagent-orchestra/\n├── agent-pulse-current/                  # Agent Pulse observability source\n├── .fleet-scorecards/                    # Ignored Fleet Scorecard overlays\n├── run-artifacts/                        # Local fleet run artifact corpus\n├── skills/SKILL.md                       # Copilot CLI skill source\n├── agents/                               # Worker, commander, and merger agent definitions\n├── schemas/                              # Bundle and collaboration record schemas\n├── bin/                                  # Terminal Stampede and Agent Orchestra launchers\n├── tests/smoke.sh                        # Local smoke gate\n├── install.sh                            # Local installer\n├── quickstart.sh                         # One-command installer\n├── scripts/activate-security.sh          # Repository security activation helper\n├── AGENTS.md                             # Contributor/agent guide\n└── archived-workflows/root/workflows/    # CI and CodeQL workflow templates\n```\n\n---\n\n## 🧪 Validate Locally\n\n```bash\nbash tests/smoke.sh\n```\n\nThe smoke test checks shell syntax, hardened runtime markers, fleet metadata,\ncommander bundle/schema integrity, Fleet Scorecard generation, Agent Pulse\nimport/poll behavior, and workflow activation safety.\n\n---\n\n## 🤝 Contributing\n\nSee [CONTRIBUTING.md](CONTRIBUTING.md). The short version:\n\n1. Keep the orchestration flow self-contained.\n2. Preserve sealed Shadow Score isolation.\n3. Preserve Fleet Scorecard seal verification and the four-question output.\n4. Use **sub-agents** in user-facing language.\n5. Keep live commentary concise.\n6. Run `bash tests/smoke.sh` before opening a PR.\n\n## License\n\nMIT — see [LICENSE](LICENSE).\n\n---\n\n🐙 Created with 💜 by [@DUBSOpenHub](https://github.com/DUBSOpenHub) with the [GitHub Copilot CLI](https://docs.github.com/copilot/concepts/agents/about-copilot-cli).\n\nLet's orchestrate! 🎼✨\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fdubsopenhub%2Fagent-orchestra","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fdubsopenhub%2Fagent-orchestra","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fdubsopenhub%2Fagent-orchestra/lists"}