# Send Colony to your agent Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually. ## Fast path - Download the package from Yavira. - Extract it into a folder your agent can access. - Paste one of the prompts below and point your agent at the extracted folder. ## Suggested prompts ### New install ```text I downloaded a skill package from Yavira. Read SKILL.md from the extracted folder and install it by following the included instructions. Tell me what you changed and call out any manual steps you could not complete. ``` ### Upgrade existing ```text I downloaded an updated skill package from Yavira. Read SKILL.md from the extracted folder, compare it with my current installation, and upgrade it while preserving any custom configuration unless the package docs explicitly say otherwise. Summarize what changed and any follow-up checks I should run. ``` ## Machine-readable fields ```json { "schemaVersion": "1.0", "item": { "slug": "multi-agent-orchestration", "name": "Colony", "source": "tencent", "type": "skill", "category": "AI 智能", "sourceUrl": "https://clawhub.ai/dabruhce/multi-agent-orchestration", "canonicalUrl": "https://clawhub.ai/dabruhce/multi-agent-orchestration", "targetPlatform": "OpenClaw" }, "install": { "downloadUrl": "/downloads/multi-agent-orchestration", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=multi-agent-orchestration", "sourcePlatform": "tencent", "targetPlatform": "OpenClaw", "packageFormat": "ZIP package", "primaryDoc": "SKILL.md", "includedAssets": [ "COLONY-DOCS.md", "package.json", "SKILL.md", "scripts/agent-wrapper.mjs", "scripts/audit.mjs", "scripts/colony-worker.mjs" ], "downloadMode": "redirect", "sourceHealth": { "source": "tencent", "slug": "multi-agent-orchestration", "status": "healthy", "reason": "direct_download_ok", "recommendedAction": "download", "checkedAt": "2026-04-30T17:14:39.506Z", "expiresAt": "2026-05-07T17:14:39.506Z", "httpStatus": 200, "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=multi-agent-orchestration", "contentType": "application/zip", "probeMethod": "head", "details": { "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=multi-agent-orchestration", "contentDisposition": "attachment; filename=\"multi-agent-orchestration-1.0.0.zip\"", "redirectLocation": null, "bodySnippet": null, "slug": "multi-agent-orchestration" }, "scope": "item", "summary": "Item download looks usable.", "detail": "Yavira can redirect you to the upstream package for this item.", "primaryActionLabel": "Download for OpenClaw", "primaryActionHref": "/downloads/multi-agent-orchestration" }, "validation": { "installChecklist": [ "Use the Yavira download entry.", "Review SKILL.md after the package is downloaded.", "Confirm the extracted package contains the expected setup assets." ], "postInstallChecks": [ "Confirm the extracted package includes the expected docs or setup files.", "Validate the skill or prompts are available in your target agent workspace.", "Capture any manual follow-up steps the agent could not complete." ] } }, "links": { "detailUrl": "https://openagent3.xyz/skills/multi-agent-orchestration", "downloadUrl": "https://openagent3.xyz/downloads/multi-agent-orchestration", "agentUrl": "https://openagent3.xyz/skills/multi-agent-orchestration/agent", "manifestUrl": "https://openagent3.xyz/skills/multi-agent-orchestration/agent.json", "briefUrl": "https://openagent3.xyz/skills/multi-agent-orchestration/agent.md" } } ``` ## Documentation ### Colony Orchestration Skill Multi-agent task delegation and process orchestration with audit logging and agent learning capabilities. ### Quick Start # Single task - auto-routed node scripts/colony.mjs dispatch "find top 5 time-series databases" # Multi-stage process node scripts/colony.mjs process validate-idea --context "AI meal planning for parents" node scripts/colony.mjs process-status # check progress node scripts/colony.mjs approve abc123 # continue past checkpoint # Check audit stats node scripts/colony.mjs audit # View agent memory node scripts/colony.mjs memory scout ### Agents AgentRoleSpecializationscuttleresearcherQuick searches, lookups, fact-findingscoutresearcherDeep market/competitor research, intelligenceforecastanalystData analysis, trends, projectionspincercoderWriting, debugging, refactoring codeshellopsGit, deployments, system tasksforgeproductPRDs, specs, roadmapsledgerfinancePricing, costs, business casesmusecreativeBrainstorming, naming, ideasscribewriterBlog posts, docs, long-form contentquillcopywriterLanding pages, sales copy, adsechosocialTweets, social posts, promotionsentryqaTesting, bug verification ### Dispatch (Auto-Route) node scripts/colony.mjs dispatch "research best practices for API rate limiting" Automatically detects the best agent based on task keywords. ### Assign to Specific Agent node scripts/colony.mjs assign scout "find top 5 time-series databases" node scripts/colony.mjs assign pincer "refactor the auth module to use JWT" node scripts/colony.mjs assign shell "deploy the staging branch" ### Check Status node scripts/colony.mjs status Shows all agents and their current tasks. ### Get Results node scripts/colony.mjs results # Latest completed task node scripts/colony.mjs results abc123 # Specific task by ID ### View History node scripts/colony.mjs history # Last 10 completed/failed node scripts/colony.mjs history --limit 20 # Custom limit ### Process Commands Processes are multi-stage workflows that chain agents together. ### List Available Processes node scripts/colony.mjs processes ### Start a Process node scripts/colony.mjs process --context "description" Examples: node scripts/colony.mjs process validate-idea --context "AI-powered meal planning for busy parents" node scripts/colony.mjs process content-pipeline --context "How to use vector databases for RAG" node scripts/colony.mjs process product-launch --context "Life Lunch ritual kit for parents" node scripts/colony.mjs process bug-triage --context "Login fails with OAuth on mobile" ### Check Process Status node scripts/colony.mjs process-status # Show latest run node scripts/colony.mjs process-status abc123 # Specific run Shows: current stage, completed stages, checkpoints, output files. ### View Process Runs node scripts/colony.mjs runs # All runs (active, paused, completed) node scripts/colony.mjs runs --limit 5 # Last 5 ### Approve Checkpoint When a process hits a checkpoint, it pauses for human approval: node scripts/colony.mjs approve abc123 Also used to retry a failed stage. ### Cancel a Process node scripts/colony.mjs cancel abc123 ### Audit Commands Track agent performance, task statistics, and system health. ### Dashboard node scripts/colony.mjs audit Shows global stats, per-agent summary, and recent events. ### Agent Details node scripts/colony.mjs audit agent scout node scripts/colony.mjs audit agent pincer Shows detailed stats for a specific agent including: Total tasks, success rate Average duration Token usage Recent failures ### Event Log node scripts/colony.mjs audit log # Last 20 events node scripts/colony.mjs audit log --limit 50 # More events ### Slowest Tasks node scripts/colony.mjs audit slow # Top 10 slowest node scripts/colony.mjs audit slow --limit 20 ### Recent Failures node scripts/colony.mjs audit failures # Last 10 failures node scripts/colony.mjs audit failures --limit 20 ### Learning Commands Agents learn from experience and share knowledge. ### Feedback Record feedback for completed tasks: node scripts/colony.mjs feedback abc123 "Great research, but needed more pricing data" ### Agent Memory Each agent has a persistent memory file with lessons learned: # View an agent's memory node scripts/colony.mjs memory scout # Add a lesson node scripts/colony.mjs memory scout add "Always check publication dates on research sources" # Add to specific sections node scripts/colony.mjs memory scout add "Use bullet points for clarity" --pattern node scripts/colony.mjs memory scout add "Missed competitor X in analysis" --mistake node scripts/colony.mjs memory scout add "Prefers markdown tables over lists" --pref ### Shared Learnings Cross-agent insights and lessons: # View all shared learnings node scripts/colony.mjs learn # Add a learning node scripts/colony.mjs learn add "validate-idea works better with 3 competitors max" --category process node scripts/colony.mjs learn add "Always verify API rate limits early" --category technical --source run-abc123 ### Global Context Shared context all agents can access: # View global context node scripts/colony.mjs context # Set preferences node scripts/colony.mjs context set preferences.codeStyle "TypeScript, functional" node scripts/colony.mjs context set preferences.timezone "America/Chicago" # Add active facts (temporary context) node scripts/colony.mjs context add-fact "We're targeting enterprise customers" node scripts/colony.mjs context add-fact "Launch deadline is Q2 2024" # Add decisions node scripts/colony.mjs context add-decision "Use Postgres over MySQL" --project "life-lunch" # Add projects node scripts/colony.mjs context add-project "life-lunch" ### Retrospective Review recent activity and generate insights: node scripts/colony.mjs retro # Last 7 days node scripts/colony.mjs retro --days 14 # Last 14 days Shows: Task completion summary Per-agent stats Failure patterns Suggested learnings ### validate-idea Validate a business idea end-to-end Stages: brainstorm → research → analyze → spec → estimate Checkpoints: after analyze Output: business-case.md ### product-launch End-to-end product launch Stages: research → spec → build → copy Checkpoints: after spec, after copy Output: market-brief.md, prd.md, code/, landing-copy.md ### content-pipeline Research, write, publish, promote content Stages: research → draft → review → publish → promote Checkpoints: review (human reviews draft) Output: research.md, draft.md, social-posts.md ### bug-triage Reproduce, fix, deploy bug fixes Stages: reproduce → fix → test → deploy Checkpoints: none (fast path) Output: bug-report.md, fix-summary.md ### customer-research Deep dive on a customer segment Stages: identify → pain-points → validate → synthesize Checkpoints: none Output: customer-profile.md, insights.md ### landing-page Create a full landing page Stages: strategy → copy → review → build Checkpoints: after copy review Output: strategy.md, copy.md, landing.html, landing.css ### How Processes Work Start - Process creates a run entry and spawns first stage agent Execute - Each stage runs with inputs from previous stages Checkpoint - If stage is a checkpoint, process pauses for approval Continue - After approval, next stage runs Complete - All stages done, outputs in colony/context// ### Context Passing {context} in task templates is replaced with your --context value Stage outputs are saved to colony/context// Next stage reads inputs from previous stage's output files Agent memory and global context are injected into prompts Full task history in tasks.json ### Parallel Stages Stages that share the same parallel_group run concurrently: stages: - id: spec agent: forge inputs: [analysis.md] parallel_group: "final" # Stages with same group run together - id: estimate agent: ledger inputs: [analysis.md] parallel_group: "final" # Same group = parallel execution When the process reaches a parallel group: All consecutive stages with the same parallel_group are collected All stages spawn concurrently (using Promise.all()) Process waits for ALL parallel stages to complete If any stage fails, the entire group fails Checkpoints work per-group (pause after all parallel stages complete) Output shows parallel execution clearly: ═══ Parallel Group: final (2 stages) ═══ → Stage 4: spec (forge) → Stage 5: estimate (ledger) --- [PARALLEL] Stage 4/5: spec --- --- [PARALLEL] Stage 5/5: estimate --- ═══ Parallel Group: final completed ═══ When to use parallel groups: Stages that read the same inputs (no dependencies on each other) Build + copy tasks (both depend on spec, not on each other) Multiple analyses of the same data Independent research threads Processes with parallel stages: validate-idea: spec + estimate run in parallel product-launch: build + copy run in parallel ### Notifications Colony can send notifications when processes hit checkpoints, complete, or fail. Notifications use openclaw cron wake to alert you. Configuration (colony/config.yaml): notifications: enabled: true # Master switch for all notifications on_checkpoint: true # Notify when process pauses at checkpoint on_complete: true # Notify when process finishes on_failure: true # Notify when process/stage fails Manage via CLI: # View current config node scripts/colony.mjs config # Disable all notifications node scripts/colony.mjs config set notifications.enabled false # Enable only failure notifications node scripts/colony.mjs config set notifications.on_checkpoint false node scripts/colony.mjs config set notifications.on_complete false node scripts/colony.mjs config set notifications.on_failure true Notification examples: 🛑 Colony checkpoint: Process "validate-idea" paused after stage "analyze". To continue: colony approve abc123 ✅ Colony complete: Process "validate-idea" finished in 120s. Run ID: abc123 ❌ Colony failed: Process "validate-idea" failed at stage "research". Error: Agent timed out. Run ID: abc123 ### Checkpoints Checkpoints pause the process for human review. Two ways to define: In process checkpoints array (after that stage completes) As a standalone stage with checkpoint: true (human-only review step) ### File Structure skills/colony/ ├── SKILL.md # This file ├── package.json # Dependencies (js-yaml) ├── colony/ │ ├── agents.yaml # Agent definitions │ ├── processes.yaml # Process definitions │ ├── config.yaml # Notification & behavior config │ ├── tasks.json # Task queue and history │ ├── runs.json # Process run tracking │ ├── feedback.json # Task feedback storage │ ├── learnings.yaml # Shared cross-agent learnings │ ├── global-context.json # Shared context for all agents │ ├── audit/ │ │ ├── log.jsonl # Append-only event log │ │ ├── global.json # Aggregate statistics │ │ └── agents/ # Per-agent statistics │ │ ├── scout.json │ │ ├── pincer.json │ │ └── ... │ ├── memory/ # Per-agent persistent memory │ │ ├── scout.md │ │ ├── pincer.md │ │ └── ... │ └── context/ # Per-task and per-run outputs │ └── / └── scripts/ ├── colony.mjs # Main CLI ├── colony-worker.mjs # Background agent executor ├── agent-wrapper.mjs # Task lifecycle utilities ├── audit.mjs # Audit system functions └── learning.mjs # Learning system functions ### Audit Events The audit log tracks these events: EventFieldstask_startedtaskId, agent, processRunId?, stage?task_completedtaskId, agent, durationMs, tokens, successtask_failedtaskId, agent, durationMs, errorcheckpoint_waitingrunId, stagecheckpoint_approvedrunId, stagecheckpoint_rejectedrunId, stage, reasonprocess_startedrunId, processId, contextprocess_completedrunId, processId, durationMsfeedback_receivedtaskId, agent, feedback ### Add New Agents Edit colony/agents.yaml: agents: myagent: role: specialist description: > What this agent does... model: anthropic/claude-sonnet-4 triggers: - keyword1 - keyword2 After adding, create their memory file: touch colony/memory/myagent.md ### Add New Processes Edit colony/processes.yaml: processes: my-process: description: "What this process does" triggers: [keyword1, keyword2] stages: - id: stage1 agent: scout task: "Do something with: {context}" outputs: [output1.md] - id: stage2 agent: pincer task: "Next step based on previous" inputs: [output1.md] outputs: [output2.md] checkpoints: [stage1] # Optional: pause after these stages ### Integration Works with OpenClaw's agent sessions. Dispatch/Assign (async): Tasks are spawned in the background and the CLI returns immediately. Use colony status to monitor progress and colony results to view output. Process stages (blocking): Multi-stage processes run sequentially, waiting for each stage to complete before proceeding. This ensures proper data flow between stages and checkpoint handling. Each agent receives: Their role description Lessons from their memory file Active facts from global context Project/preference context ### Validate a Startup Idea node scripts/colony.mjs process validate-idea \\ --context "Subscription box for home coffee brewing experiments" Watch as it flows: brainstorm → research → analyze → (checkpoint) → spec → estimate ### Write and Publish a Blog Post node scripts/colony.mjs process content-pipeline \\ --context "Why RAG is eating traditional search" Stages: research → draft → (human review) → publish → promote ### Quick Research Task node scripts/colony.mjs dispatch "compare Pinecone vs Weaviate vs Milvus" Auto-routes to scout, returns comparison. ### Track Performance # After several tasks, check overall health node scripts/colony.mjs audit # Deep dive into a struggling agent node scripts/colony.mjs audit agent pincer node scripts/colony.mjs audit failures # Add learnings from issues node scripts/colony.mjs memory pincer add "Handle file not found errors gracefully" --mistake ## Trust - Source: tencent - Verification: Indexed source record - Publisher: dabruhce - Version: 1.0.0 ## Source health - Status: healthy - Item download looks usable. - Yavira can redirect you to the upstream package for this item. - Health scope: item - Reason: direct_download_ok - Checked at: 2026-04-30T17:14:39.506Z - Expires at: 2026-05-07T17:14:39.506Z - Recommended action: Download for OpenClaw ## Links - [Detail page](https://openagent3.xyz/skills/multi-agent-orchestration) - [Send to Agent page](https://openagent3.xyz/skills/multi-agent-orchestration/agent) - [JSON manifest](https://openagent3.xyz/skills/multi-agent-orchestration/agent.json) - [Markdown brief](https://openagent3.xyz/skills/multi-agent-orchestration/agent.md) - [Download page](https://openagent3.xyz/downloads/multi-agent-orchestration)