{
  "schemaVersion": "1.0",
  "item": {
    "slug": "lattice",
    "name": "Lattice",
    "source": "tencent",
    "type": "skill",
    "category": "AI 智能",
    "sourceUrl": "https://clawhub.ai/CNF6682/lattice",
    "canonicalUrl": "https://clawhub.ai/CNF6682/lattice",
    "targetPlatform": "OpenClaw"
  },
  "install": {
    "downloadMode": "redirect",
    "downloadUrl": "/downloads/lattice",
    "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=lattice",
    "sourcePlatform": "tencent",
    "targetPlatform": "OpenClaw",
    "installMethod": "Manual import",
    "extraction": "Extract archive",
    "prerequisites": [
      "OpenClaw"
    ],
    "packageFormat": "ZIP package",
    "includedAssets": [
      "SKILL.md",
      "templates/ORG/ASSET_REGISTRY.md",
      "templates/ORG/DEPARTMENTS/example-dept/CHARTER.md",
      "templates/ORG/DEPARTMENTS/example-dept/HANDOFF.md",
      "templates/ORG/DEPARTMENTS/example-dept/RUNBOOK.md",
      "templates/ORG/DEPARTMENTS/example-dept/STATE.json"
    ],
    "primaryDoc": "SKILL.md",
    "quickSetup": [
      "Download the package from Yavira.",
      "Extract the archive and review SKILL.md first.",
      "Import or place the package into your OpenClaw setup."
    ],
    "agentAssist": {
      "summary": "Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.",
      "steps": [
        "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."
      ],
      "prompts": [
        {
          "label": "New install",
          "body": "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."
        },
        {
          "label": "Upgrade existing",
          "body": "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."
        }
      ]
    },
    "sourceHealth": {
      "source": "tencent",
      "status": "healthy",
      "reason": "direct_download_ok",
      "recommendedAction": "download",
      "checkedAt": "2026-04-30T16:55:25.780Z",
      "expiresAt": "2026-05-07T16:55:25.780Z",
      "httpStatus": 200,
      "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=network",
      "contentType": "application/zip",
      "probeMethod": "head",
      "details": {
        "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=network",
        "contentDisposition": "attachment; filename=\"network-1.0.0.zip\"",
        "redirectLocation": null,
        "bodySnippet": null
      },
      "scope": "source",
      "summary": "Source download looks usable.",
      "detail": "Yavira can redirect you to the upstream package for this source.",
      "primaryActionLabel": "Download for OpenClaw",
      "primaryActionHref": "/downloads/lattice"
    },
    "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."
      ]
    },
    "downloadPageUrl": "https://openagent3.xyz/downloads/lattice",
    "agentPageUrl": "https://openagent3.xyz/skills/lattice/agent",
    "manifestUrl": "https://openagent3.xyz/skills/lattice/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/lattice/agent.md"
  },
  "agentAssist": {
    "summary": "Hand the extracted package to your coding agent with a concrete install brief instead of figuring it out manually.",
    "steps": [
      "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."
    ],
    "prompts": [
      {
        "label": "New install",
        "body": "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."
      },
      {
        "label": "Upgrade existing",
        "body": "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."
      }
    ]
  },
  "documentation": {
    "source": "clawhub",
    "primaryDoc": "SKILL.md",
    "sections": [
      {
        "title": "Lattice",
        "body": "Lattice is a file-based operating system for AI agent teams. It replaces volatile chat windows with persistent files, enabling agents to work stably through long, iterative development cycles via an 8-phase execution pipeline.\n\nWhy Lattice?\n\nStable long-running execution — File-driven state machine ensures agents stay on track across sessions. No context window overflow, no drift. Tasks complete reliably over hours or days through structured iteration.\nThree-tier failure handling — When an agent gets stuck: (1) Model Escalation retries with stronger models, (2) Peer Consult gathers parallel opinions from multiple models, (3) Auto-Triage makes a judgment call (relax constraints / defer / block for human). Most blockers resolve automatically.\nPer-phase model configuration — Thinking-heavy phases (planning, review, architecture) need strong reasoning models; token-heavy phases (implementation, testing) can use cost-efficient coding models. This dramatically reduces overall token cost without sacrificing quality where it matters.\n\nExample model assignment (from a production setup):\n\nPhaseModel tierExampleConstitute (architecture)Strong reasoningClaude OpusResearchStrong reasoningGemini ProSpecify (design)Strong reasoningClaude OpusPlan + TasksStrong reasoningClaude OpusImplementCost-efficient codingGPT CodexTestCost-efficient codingGPT CodexReviewStrong reasoningClaude OpusGap AnalysisStrong reasoningGemini Pro\n\nThe key insight: implementation and testing consume the most tokens (writing/running code), but don't require the most expensive models. Planning and review consume fewer tokens but need deeper reasoning. Match model strength to cognitive demand, not token volume.\n\nMulti-project parallel execution — Run multiple projects simultaneously, each with its own cron-scheduled orchestrator. Combined with OpenClaw's cron system, projects advance autonomously on independent cadences.\n\nTemplates are bundled at templates/ORG/ relative to this skill directory."
      },
      {
        "title": "Quick Reference",
        "body": "Full design doc: templates/ORG/PROJECTS/pipeline-framework/DESIGN.md\nPipeline guide (all agents): templates/ORG/PIPELINE_GUIDE.md\nSub-agent guide: templates/ORG/PROJECTS/pipeline-framework/templates/PIPELINE_GUIDE_FOR_SUBAGENTS.md\nOrchestrator prompt template: templates/ORG/PROJECTS/pipeline-framework/templates/ORCHESTRATOR_PROMPT.template.md\nPhase prompt templates: templates/ORG/PROJECTS/pipeline-framework/templates/PHASE_PROMPTS/\nState machine template: templates/ORG/PROJECTS/pipeline-framework/templates/PIPELINE_STATE.template.json"
      },
      {
        "title": "1. Gather Information",
        "body": "Ask the user for:\n\nOrganization name (e.g. \"Acme Labs\")\nTarget directory — where to create the ORG/ folder (default: current workspace root)\nDepartments — list of department names (e.g. Research, Engineering, Reliability)\nFirst project (optional) — name + one-line description\n\nKeep it conversational. Don't dump all questions at once."
      },
      {
        "title": "2. Scaffold the ORG Directory",
        "body": "Copy the entire templates/ORG/ directory to <target>/ORG/.\n\nThen customize:\n\nORG_README.md — Replace the example org structure in §5 with the user's actual departments\nTASKBOARD.md — Leave the template structure intact (user fills in priorities later)\nDepartments — For each department the user listed:\n\nCopy DEPARTMENTS/example-dept/ → DEPARTMENTS/<dept-name>/\nIn each copied department, replace <Department Name> placeholders in CHARTER.md, RUNBOOK.md, HANDOFF.md with the actual department name\nReset STATE.json to {\"lastRun\": null, \"cursor\": null, \"notes\": \"Initial state\"}\n\n\nRemove DEPARTMENTS/example-dept/ after creating real departments (unless user wants to keep it as reference)"
      },
      {
        "title": "3. Create the First Project (if requested)",
        "body": "Copy PROJECTS/example-project/ → PROJECTS/<project-name>/\nUpdate STATUS.md with the project name\nUpdate DECISIONS.md header\nConfigure PIPELINE_STATE.json (see \"Configure Pipeline State\" below)\nRemove PROJECTS/example-project/ after creating the real project"
      },
      {
        "title": "4. Configure Pipeline State",
        "body": "Read templates/ORG/PROJECTS/pipeline-framework/templates/PIPELINE_STATE.template.json as the reference.\n\nAsk the user:\n\nWhich agents will run each phase? (agentId per role — or a single agent for all)\nWhich models for each phase? (or a default model)\nEscalation chain — list of models from cheapest to strongest (e.g. [\"gflash\", \"gpro\", \"sonnet\"])\nPeer consult models — which models to consult in parallel when stuck\nSynthesizer/triage model — typically the strongest available model\nNotification channel (optional) — where to send pipeline status updates\n\nFill in the project's PIPELINE_STATE.json with these values, replacing all <placeholder> tokens."
      },
      {
        "title": "5. Set Up the Orchestrator Cron Job",
        "body": "Read templates/ORG/PROJECTS/pipeline-framework/templates/ORCHESTRATOR_PROMPT.template.md.\n\nCreate a cron job using the cron tool:\n\nSchedule: every 30 minutes (adjustable)\nSession target: isolated\nPayload kind: agentTurn\nModel: the user's chosen orchestrator model\nMessage: the orchestrator prompt template, with all <placeholder> tokens filled in:\n\n<project> → project name\n<org-root> → absolute path to the ORG directory\n<project-root> → absolute path to the project directory\n<repo-root> → absolute path to the code repository (ask user)\nPhase prompt paths → absolute paths to the skill's bundled phase prompt templates\n\nTell the user the cron job ID so they can manage it later."
      },
      {
        "title": "6. Summary",
        "body": "Print a brief summary:\n\nORG directory location\nDepartments created\nProject(s) created\nCron job ID and schedule\nRemind them to fill in TASKBOARD.md with initial priorities"
      },
      {
        "title": "Task: Add a New Project (lattice new-project)",
        "body": "Ask for: project name, one-line description, code repo path\nCopy templates/ORG/PROJECTS/example-project/ → ORG/PROJECTS/<name>/\nCustomize STATUS.md, DECISIONS.md, PIPELINE_STATE.json (same as step 3-4 above)\nOptionally create a new orchestrator cron job for this project"
      },
      {
        "title": "Task: Add a New Department (lattice new-dept)",
        "body": "Ask for: department name, mission (one sentence)\nCopy templates/ORG/DEPARTMENTS/example-dept/ → ORG/DEPARTMENTS/<name>/\nFill in CHARTER.md with the department name and mission\nUpdate ORG_README.md §5 to include the new department"
      },
      {
        "title": "Task: Check Organization Status (lattice status)",
        "body": "Read ORG/TASKBOARD.md — summarize active priorities\nFor each project in ORG/PROJECTS/:\n\nRead STATUS.md — current phase and progress\nRead PIPELINE_STATE.json — phase statuses, blockers, run number\n\n\nFor each department in ORG/DEPARTMENTS/:\n\nRead HANDOFF.md — current state and blockers\n\n\nPresent a concise status report"
      },
      {
        "title": "8 Phases",
        "body": "Constitute → Research → Specify → Plan+Tasks → Implement → Test → Review → Gap Analysis"
      },
      {
        "title": "3-Layer Assistance (when a phase gets stuck)",
        "body": "Model Escalation — retry with progressively stronger models\nPeer Consult — parallel multi-model consultation + synthesis\nAuto-Triage — automated judge decides: RELAX (loosen constraints) / DEFER (next iteration) / BLOCK (wait for human)"
      },
      {
        "title": "Key Files per Project",
        "body": "ORG/PROJECTS/<project>/\n├── STATUS.md              # Human-readable status\n├── DECISIONS.md           # Key decisions + rationale\n├── PIPELINE_STATE.json    # Phase state machine\n├── PIPELINE_LOG.jsonl     # Append-only history\n├── pipeline/              # Current run artifacts\n└── pipeline_archive/      # Historical runs"
      }
    ],
    "body": "Lattice\n\nLattice is a file-based operating system for AI agent teams. It replaces volatile chat windows with persistent files, enabling agents to work stably through long, iterative development cycles via an 8-phase execution pipeline.\n\nWhy Lattice?\n\nStable long-running execution — File-driven state machine ensures agents stay on track across sessions. No context window overflow, no drift. Tasks complete reliably over hours or days through structured iteration.\nThree-tier failure handling — When an agent gets stuck: (1) Model Escalation retries with stronger models, (2) Peer Consult gathers parallel opinions from multiple models, (3) Auto-Triage makes a judgment call (relax constraints / defer / block for human). Most blockers resolve automatically.\nPer-phase model configuration — Thinking-heavy phases (planning, review, architecture) need strong reasoning models; token-heavy phases (implementation, testing) can use cost-efficient coding models. This dramatically reduces overall token cost without sacrificing quality where it matters.\n\nExample model assignment (from a production setup):\n\nPhase\tModel tier\tExample\nConstitute (architecture)\tStrong reasoning\tClaude Opus\nResearch\tStrong reasoning\tGemini Pro\nSpecify (design)\tStrong reasoning\tClaude Opus\nPlan + Tasks\tStrong reasoning\tClaude Opus\nImplement\tCost-efficient coding\tGPT Codex\nTest\tCost-efficient coding\tGPT Codex\nReview\tStrong reasoning\tClaude Opus\nGap Analysis\tStrong reasoning\tGemini Pro\n\nThe key insight: implementation and testing consume the most tokens (writing/running code), but don't require the most expensive models. Planning and review consume fewer tokens but need deeper reasoning. Match model strength to cognitive demand, not token volume.\n\nMulti-project parallel execution — Run multiple projects simultaneously, each with its own cron-scheduled orchestrator. Combined with OpenClaw's cron system, projects advance autonomously on independent cadences.\n\nTemplates are bundled at templates/ORG/ relative to this skill directory.\n\nQuick Reference\nFull design doc: templates/ORG/PROJECTS/pipeline-framework/DESIGN.md\nPipeline guide (all agents): templates/ORG/PIPELINE_GUIDE.md\nSub-agent guide: templates/ORG/PROJECTS/pipeline-framework/templates/PIPELINE_GUIDE_FOR_SUBAGENTS.md\nOrchestrator prompt template: templates/ORG/PROJECTS/pipeline-framework/templates/ORCHESTRATOR_PROMPT.template.md\nPhase prompt templates: templates/ORG/PROJECTS/pipeline-framework/templates/PHASE_PROMPTS/\nState machine template: templates/ORG/PROJECTS/pipeline-framework/templates/PIPELINE_STATE.template.json\nTask: Initialize a New Organization (lattice init)\n1. Gather Information\n\nAsk the user for:\n\nOrganization name (e.g. \"Acme Labs\")\nTarget directory — where to create the ORG/ folder (default: current workspace root)\nDepartments — list of department names (e.g. Research, Engineering, Reliability)\nFirst project (optional) — name + one-line description\n\nKeep it conversational. Don't dump all questions at once.\n\n2. Scaffold the ORG Directory\n\nCopy the entire templates/ORG/ directory to <target>/ORG/.\n\nThen customize:\n\nORG_README.md — Replace the example org structure in §5 with the user's actual departments\nTASKBOARD.md — Leave the template structure intact (user fills in priorities later)\nDepartments — For each department the user listed:\nCopy DEPARTMENTS/example-dept/ → DEPARTMENTS/<dept-name>/\nIn each copied department, replace <Department Name> placeholders in CHARTER.md, RUNBOOK.md, HANDOFF.md with the actual department name\nReset STATE.json to {\"lastRun\": null, \"cursor\": null, \"notes\": \"Initial state\"}\nRemove DEPARTMENTS/example-dept/ after creating real departments (unless user wants to keep it as reference)\n3. Create the First Project (if requested)\nCopy PROJECTS/example-project/ → PROJECTS/<project-name>/\nUpdate STATUS.md with the project name\nUpdate DECISIONS.md header\nConfigure PIPELINE_STATE.json (see \"Configure Pipeline State\" below)\nRemove PROJECTS/example-project/ after creating the real project\n4. Configure Pipeline State\n\nRead templates/ORG/PROJECTS/pipeline-framework/templates/PIPELINE_STATE.template.json as the reference.\n\nAsk the user:\n\nWhich agents will run each phase? (agentId per role — or a single agent for all)\nWhich models for each phase? (or a default model)\nEscalation chain — list of models from cheapest to strongest (e.g. [\"gflash\", \"gpro\", \"sonnet\"])\nPeer consult models — which models to consult in parallel when stuck\nSynthesizer/triage model — typically the strongest available model\nNotification channel (optional) — where to send pipeline status updates\n\nFill in the project's PIPELINE_STATE.json with these values, replacing all <placeholder> tokens.\n\n5. Set Up the Orchestrator Cron Job\n\nRead templates/ORG/PROJECTS/pipeline-framework/templates/ORCHESTRATOR_PROMPT.template.md.\n\nCreate a cron job using the cron tool:\n\nSchedule: every 30 minutes (adjustable)\nSession target: isolated\nPayload kind: agentTurn\nModel: the user's chosen orchestrator model\nMessage: the orchestrator prompt template, with all <placeholder> tokens filled in:\n<project> → project name\n<org-root> → absolute path to the ORG directory\n<project-root> → absolute path to the project directory\n<repo-root> → absolute path to the code repository (ask user)\nPhase prompt paths → absolute paths to the skill's bundled phase prompt templates\n\nTell the user the cron job ID so they can manage it later.\n\n6. Summary\n\nPrint a brief summary:\n\nORG directory location\nDepartments created\nProject(s) created\nCron job ID and schedule\nRemind them to fill in TASKBOARD.md with initial priorities\nTask: Add a New Project (lattice new-project)\nAsk for: project name, one-line description, code repo path\nCopy templates/ORG/PROJECTS/example-project/ → ORG/PROJECTS/<name>/\nCustomize STATUS.md, DECISIONS.md, PIPELINE_STATE.json (same as step 3-4 above)\nOptionally create a new orchestrator cron job for this project\nTask: Add a New Department (lattice new-dept)\nAsk for: department name, mission (one sentence)\nCopy templates/ORG/DEPARTMENTS/example-dept/ → ORG/DEPARTMENTS/<name>/\nFill in CHARTER.md with the department name and mission\nUpdate ORG_README.md §5 to include the new department\nTask: Check Organization Status (lattice status)\nRead ORG/TASKBOARD.md — summarize active priorities\nFor each project in ORG/PROJECTS/:\nRead STATUS.md — current phase and progress\nRead PIPELINE_STATE.json — phase statuses, blockers, run number\nFor each department in ORG/DEPARTMENTS/:\nRead HANDOFF.md — current state and blockers\nPresent a concise status report\nPipeline Architecture (for reference)\n8 Phases\nConstitute → Research → Specify → Plan+Tasks → Implement → Test → Review → Gap Analysis\n\n3-Layer Assistance (when a phase gets stuck)\nModel Escalation — retry with progressively stronger models\nPeer Consult — parallel multi-model consultation + synthesis\nAuto-Triage — automated judge decides: RELAX (loosen constraints) / DEFER (next iteration) / BLOCK (wait for human)\nKey Files per Project\nORG/PROJECTS/<project>/\n├── STATUS.md              # Human-readable status\n├── DECISIONS.md           # Key decisions + rationale\n├── PIPELINE_STATE.json    # Phase state machine\n├── PIPELINE_LOG.jsonl     # Append-only history\n├── pipeline/              # Current run artifacts\n└── pipeline_archive/      # Historical runs"
  },
  "trust": {
    "sourceLabel": "tencent",
    "provenanceUrl": "https://clawhub.ai/CNF6682/lattice",
    "publisherUrl": "https://clawhub.ai/CNF6682/lattice",
    "owner": "CNF6682",
    "version": "1.2.1",
    "license": null,
    "verificationStatus": "Indexed source record"
  },
  "links": {
    "detailUrl": "https://openagent3.xyz/skills/lattice",
    "downloadUrl": "https://openagent3.xyz/downloads/lattice",
    "agentUrl": "https://openagent3.xyz/skills/lattice/agent",
    "manifestUrl": "https://openagent3.xyz/skills/lattice/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/lattice/agent.md"
  }
}