{
  "schemaVersion": "1.0",
  "item": {
    "slug": "consilium",
    "name": "Consilium",
    "source": "tencent",
    "type": "skill",
    "category": "AI 智能",
    "sourceUrl": "https://clawhub.ai/morozsm/consilium",
    "canonicalUrl": "https://clawhub.ai/morozsm/consilium",
    "targetPlatform": "OpenClaw"
  },
  "install": {
    "downloadMode": "redirect",
    "downloadUrl": "/downloads/consilium",
    "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=consilium",
    "sourcePlatform": "tencent",
    "targetPlatform": "OpenClaw",
    "installMethod": "Manual import",
    "extraction": "Extract archive",
    "prerequisites": [
      "OpenClaw"
    ],
    "packageFormat": "ZIP package",
    "includedAssets": [
      "ROADMAP.md",
      "SKILL.md",
      "references/PROTOCOL.md"
    ],
    "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/consilium"
    },
    "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/consilium",
    "agentPageUrl": "https://openagent3.xyz/skills/consilium/agent",
    "manifestUrl": "https://openagent3.xyz/skills/consilium/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/consilium/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": "Consilium — True Multi-Model Deliberation",
        "body": "Ask a hard question → 3-5 AI models from different providers analyze it independently → you get a synthesis with consensus, disagreements, action items, and minority opinions.\n\nUnlike other council skills: this uses genuinely different models (Anthropic + OpenAI + Google + others), not one model playing multiple roles. Different training data = different blind spots = better coverage.\n\nAlways respond in the same language as the user's question."
      },
      {
        "title": "Examples",
        "body": "/council Should we migrate from monolith to microservices given our 4-person team?\n/council --profile fast Evaluate the risks of this investment strategy\n/council How to resolve a complex equity dispute with my co-founder?\nAfter results: \"Tell me more about what Gemini said on point 3\" (follow-up with specific panelist)"
      },
      {
        "title": "Requirements",
        "body": "Minimum 3 models from different providers in agents.defaults.models allowlist\nTools: sessions_spawn, subagents, sessions_history (enabled by default)\nEach council run = 3-5 API calls (one per model) + synthesis\nNo additional API keys, Python scripts, or external dependencies"
      },
      {
        "title": "Privacy & Data",
        "body": "Your question is sent to each model provider in your panel. Only use models/providers you trust.\ncouncil-panel.json (saved to workspace root) contains only model names and slot assignments, not queries or responses.\nPanelist responses exist only in sub-agent session memory and are auto-archived per your OpenClaw settings.\nNo data is sent to external services beyond your configured model providers."
      },
      {
        "title": "Panel",
        "body": "On first use, check available models and ask the user to confirm the panel. Save to workspace root as council-panel.json for reuse. User can re-run panel selection anytime with --models."
      },
      {
        "title": "Slot roles (fill from available models)",
        "body": "SlotRoleGood candidatesDeep thinkerNuance, system thinkingClaude Opus, GPT-5, Gemini ProPragmatistConcise, actionableClaude Sonnet, GPT-mini, Gemini FlashBroad analystWide knowledge, structureGPT-5, Gemini Pro, Claude OpusTechnicalRigor, edge casesGemini Pro, Claude Sonnet, GLMContrarianChallenge assumptionsGLM, any model with contrarian lens\n\nRules: Each slot = different model. Prefer different providers. Min 3 models to run. If fewer than 3 available, inform user."
      },
      {
        "title": "Example council-panel.json",
        "body": "{\n  \"panel\": [\n    { \"slot\": \"deep_thinker\", \"model\": \"anthropic/claude-opus-4-6\", \"lens\": \"Deep analysis\" },\n    { \"slot\": \"pragmatist\", \"model\": \"anthropic/claude-sonnet-4-5\", \"lens\": \"Pragmatic\" },\n    { \"slot\": \"broad_analyst\", \"model\": \"github-copilot/gpt-5.2\", \"lens\": \"Broad knowledge\" }\n  ],\n  \"confirmed\": \"2026-02-24\"\n}"
      },
      {
        "title": "Profiles",
        "body": "thorough (default): All panel slots, quorum = max(slots - 2, 2)\nbalanced: 3 strongest slots, quorum 2\nfast: 2 fastest slots, quorum 2"
      },
      {
        "title": "Workflow",
        "body": "Dispatch — spawn panelists in parallel (sessions_spawn, mode=run, timeout 120s). Assign unique lens per slot. Detect question language, hardcode in prompt. Tell user: \"Panel dispatched, ~60s. Send a follow-up when ready.\"\nCollect — on user's follow-up: subagents list → sessions_history. Synthesize when quorum met.\nDebate (only if --rounds 2) — anonymized digest → rebuttals. See references/PROTOCOL.md.\nSynthesize — produce output below."
      },
      {
        "title": "Output Format",
        "body": "## Council of Experts\n**Question:** ... | **Panel:** ... | **Profile:** ...\n---\n### Positions\n**{Model}** ({lens}) — {2-3 sentence summary}\n\n### ✅ Consensus\n### ⚡ Disagreements\n### 🗣️ Minority opinions\n\n### 🎯 Synthesis\nAgreement: 🟢 strong (4-5) | 🟡 mixed (3) | 🔴 split\n\n### 📋 Action Items\n1. **{Highest priority}** — {effort/time estimate}\n2. **{Next action}** — {estimate}\n3. **{Next action}** — {estimate}\n\nRandomize position order. Quote with attribution. Preserve minority views. Never fabricate consensus. Section headers and content in user's language."
      },
      {
        "title": "Follow-up",
        "body": "After synthesis, the user can drill deeper with a specific panelist:\n\n\"Tell me more about what GPT said on point 2\"\n\"I want the contrarian's take on the action items\"\n\nUse sessions_history to retrieve that panelist's full response, then expand on the specific point in that model's perspective."
      },
      {
        "title": "Flags",
        "body": "--profile thorough|balanced|fast · --models <list> · --skip <model> · --rounds 2 · --quorum N · --timeout N · --lens \"...\" · --lenses \"a,b,c\"\n\nPrompt templates, debate mechanics, error handling → references/PROTOCOL.md"
      }
    ],
    "body": "Consilium — True Multi-Model Deliberation\n\nAsk a hard question → 3-5 AI models from different providers analyze it independently → you get a synthesis with consensus, disagreements, action items, and minority opinions.\n\nUnlike other council skills: this uses genuinely different models (Anthropic + OpenAI + Google + others), not one model playing multiple roles. Different training data = different blind spots = better coverage.\n\nAlways respond in the same language as the user's question.\n\nExamples\n/council Should we migrate from monolith to microservices given our 4-person team?\n/council --profile fast Evaluate the risks of this investment strategy\n/council How to resolve a complex equity dispute with my co-founder?\nAfter results: \"Tell me more about what Gemini said on point 3\" (follow-up with specific panelist)\nRequirements\nMinimum 3 models from different providers in agents.defaults.models allowlist\nTools: sessions_spawn, subagents, sessions_history (enabled by default)\nEach council run = 3-5 API calls (one per model) + synthesis\nNo additional API keys, Python scripts, or external dependencies\nPrivacy & Data\nYour question is sent to each model provider in your panel. Only use models/providers you trust.\ncouncil-panel.json (saved to workspace root) contains only model names and slot assignments, not queries or responses.\nPanelist responses exist only in sub-agent session memory and are auto-archived per your OpenClaw settings.\nNo data is sent to external services beyond your configured model providers.\nPanel\n\nOn first use, check available models and ask the user to confirm the panel. Save to workspace root as council-panel.json for reuse. User can re-run panel selection anytime with --models.\n\nSlot roles (fill from available models)\nSlot\tRole\tGood candidates\nDeep thinker\tNuance, system thinking\tClaude Opus, GPT-5, Gemini Pro\nPragmatist\tConcise, actionable\tClaude Sonnet, GPT-mini, Gemini Flash\nBroad analyst\tWide knowledge, structure\tGPT-5, Gemini Pro, Claude Opus\nTechnical\tRigor, edge cases\tGemini Pro, Claude Sonnet, GLM\nContrarian\tChallenge assumptions\tGLM, any model with contrarian lens\n\nRules: Each slot = different model. Prefer different providers. Min 3 models to run. If fewer than 3 available, inform user.\n\nExample council-panel.json\n{\n  \"panel\": [\n    { \"slot\": \"deep_thinker\", \"model\": \"anthropic/claude-opus-4-6\", \"lens\": \"Deep analysis\" },\n    { \"slot\": \"pragmatist\", \"model\": \"anthropic/claude-sonnet-4-5\", \"lens\": \"Pragmatic\" },\n    { \"slot\": \"broad_analyst\", \"model\": \"github-copilot/gpt-5.2\", \"lens\": \"Broad knowledge\" }\n  ],\n  \"confirmed\": \"2026-02-24\"\n}\n\nProfiles\nthorough (default): All panel slots, quorum = max(slots - 2, 2)\nbalanced: 3 strongest slots, quorum 2\nfast: 2 fastest slots, quorum 2\nWorkflow\nDispatch — spawn panelists in parallel (sessions_spawn, mode=run, timeout 120s). Assign unique lens per slot. Detect question language, hardcode in prompt. Tell user: \"Panel dispatched, ~60s. Send a follow-up when ready.\"\nCollect — on user's follow-up: subagents list → sessions_history. Synthesize when quorum met.\nDebate (only if --rounds 2) — anonymized digest → rebuttals. See references/PROTOCOL.md.\nSynthesize — produce output below.\nOutput Format\n## Council of Experts\n**Question:** ... | **Panel:** ... | **Profile:** ...\n---\n### Positions\n**{Model}** ({lens}) — {2-3 sentence summary}\n\n### ✅ Consensus\n### ⚡ Disagreements\n### 🗣️ Minority opinions\n\n### 🎯 Synthesis\nAgreement: 🟢 strong (4-5) | 🟡 mixed (3) | 🔴 split\n\n### 📋 Action Items\n1. **{Highest priority}** — {effort/time estimate}\n2. **{Next action}** — {estimate}\n3. **{Next action}** — {estimate}\n\n\nRandomize position order. Quote with attribution. Preserve minority views. Never fabricate consensus. Section headers and content in user's language.\n\nFollow-up\n\nAfter synthesis, the user can drill deeper with a specific panelist:\n\n\"Tell me more about what GPT said on point 2\"\n\"I want the contrarian's take on the action items\"\n\nUse sessions_history to retrieve that panelist's full response, then expand on the specific point in that model's perspective.\n\nFlags\n\n--profile thorough|balanced|fast · --models <list> · --skip <model> · --rounds 2 · --quorum N · --timeout N · --lens \"...\" · --lenses \"a,b,c\"\n\nPrompt templates, debate mechanics, error handling → references/PROTOCOL.md"
  },
  "trust": {
    "sourceLabel": "tencent",
    "provenanceUrl": "https://clawhub.ai/morozsm/consilium",
    "publisherUrl": "https://clawhub.ai/morozsm/consilium",
    "owner": "morozsm",
    "version": "1.1.0",
    "license": null,
    "verificationStatus": "Indexed source record"
  },
  "links": {
    "detailUrl": "https://openagent3.xyz/skills/consilium",
    "downloadUrl": "https://openagent3.xyz/downloads/consilium",
    "agentUrl": "https://openagent3.xyz/skills/consilium/agent",
    "manifestUrl": "https://openagent3.xyz/skills/consilium/agent.json",
    "briefUrl": "https://openagent3.xyz/skills/consilium/agent.md"
  }
}