# Send War Room 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. Then review README.md for any prerequisites, environment setup, or post-install checks. 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. Then review README.md for any prerequisites, environment setup, or post-install checks. Summarize what changed and any follow-up checks I should run. ``` ## Machine-readable fields ```json { "schemaVersion": "1.0", "item": { "slug": "war-room", "name": "War Room", "source": "tencent", "type": "skill", "category": "AI 智能", "sourceUrl": "https://clawhub.ai/maxkle1nz/war-room", "canonicalUrl": "https://clawhub.ai/maxkle1nz/war-room", "targetPlatform": "OpenClaw" }, "install": { "downloadUrl": "/downloads/war-room", "sourceDownloadUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=war-room", "sourcePlatform": "tencent", "targetPlatform": "OpenClaw", "packageFormat": "ZIP package", "primaryDoc": "SKILL.md", "includedAssets": [ "README.md", "SKILL.md", "references/agent-roles.md", "references/dna-template.md", "references/wave-protocol.md", "scripts/init_war_room.sh" ], "downloadMode": "redirect", "sourceHealth": { "source": "tencent", "slug": "war-room", "status": "healthy", "reason": "direct_download_ok", "recommendedAction": "download", "checkedAt": "2026-05-04T11:45:27.954Z", "expiresAt": "2026-05-11T11:45:27.954Z", "httpStatus": 200, "finalUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=war-room", "contentType": "application/zip", "probeMethod": "head", "details": { "probeUrl": "https://wry-manatee-359.convex.site/api/v1/download?slug=war-room", "contentDisposition": "attachment; filename=\"war-room-1.1.0.zip\"", "redirectLocation": null, "bodySnippet": null, "slug": "war-room" }, "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/war-room" }, "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/war-room", "downloadUrl": "https://openagent3.xyz/downloads/war-room", "agentUrl": "https://openagent3.xyz/skills/war-room/agent", "manifestUrl": "https://openagent3.xyz/skills/war-room/agent.json", "briefUrl": "https://openagent3.xyz/skills/war-room/agent.md" } } ``` ## Documentation ### War Room A methodology for running multi-agent brainstorming and execution sessions. Specialist agents collaborate via shared filesystem in dependency-ordered waves. A CHAOS agent (devil's advocate) shadows every wave. Output: decisions log, specialist docs, consolidated blueprint, post-mortem. ### Quick Start Initialize: Run bash skills/war-room/scripts/init_war_room.sh to create the project folder structure under war-rooms//. Brief: Fill in war-rooms//BRIEF.md with the project description, goals, constraints, and known risks. Inject DNA: Copy skills/war-room/references/dna-template.md → war-rooms//DNA.md. Customize if needed (add project-specific identity, owner name). Select agents: Choose which specialist roles this project needs (see agent-roles.md). Not every project needs all roles. Run waves: Execute the wave protocol below. Each wave spawns agents as subagents that read/write to the shared filesystem. Consolidate: Merge all agent outputs into a blueprint in war-rooms//artifacts/. Post-mortem: Write lessons to war-rooms//lessons/. ### The Wave Protocol Full protocol details: wave-protocol.md ### Wave 0: Prove It (mandatory) Before any spec work, identify the single riskiest assumption and test it with real work (code spike, prototype, market research, etc.). 30 min max. If it fails, pivot BEFORE spending tokens on detailed specs. ### Waves 1–N: Specialist Execution Each wave deploys a group of agents that can work in parallel (no inter-dependencies within a wave). Agents in later waves depend on earlier waves' outputs. Planning a wave: List all agents needed for the project Build a dependency graph (who needs whose output?) Group agents with no mutual dependencies into the same wave Order waves by dependency Each agent in a wave: Reads: BRIEF.md, DNA.md, DECISIONS.md, and any prior agents' output folders Writes: To agents// — their specs, findings, decisions Updates: DECISIONS.md (their domain decisions), STATUS.md (their completion status) Communicates: Via comms/ for cross-agent questions/challenges Spawning agents: Each agent is a subagent. Its system prompt includes: The DNA (from DNA.md) Its role briefing (from agent-roles.md) The project brief Instruction to read prior wave outputs and write to its own folder ### Pivot Gate (between every wave) Before launching each new wave, ask: "Has any fundamental assumption changed since the last wave?" If YES → affected agents from prior waves must re-evaluate. Mark voided decisions as **VOIDED** in DECISIONS.md. If NO → proceed. ### CHAOS Shadows Every Wave CHAOS is not a separate wave — it shadows all waves. After each wave completes, CHAOS: Reads every agent's output from that wave Files challenges to agents/chaos/challenges.md Format: [C-ID] CHALLENGE to D### — attack — verdict (SURVIVE/WOUNDED/KILLED) WOUNDED = valid concern, needs mitigation. KILLED = decision must be reversed. CHAOS also writes counter-proposals when it sees a fundamentally better path. ### Consolidation Wave (final) One agent (or the orchestrator) merges all specialist outputs into a single blueprint: Read all agents/*/ outputs Resolve contradictions (flag any that remain) Produce unified document in artifacts/-BLUEPRINT.md Include: architecture, scope, risks, roadmap, via negativa (what's NOT included) CHAOS reviews the blueprint for internal contradictions ### Post-Mortem After consolidation, write lessons/session-N-postmortem.md: What went well What went wrong (wasted work, late catches, process failures) Root causes Lessons for next session ### Agent Selection Guide Not every project needs every role. Match roles to scope: Project TypeTypical AgentsSoftware MVPARCH, PM, DEV, UX, SEC, QA, CHAOSBusiness strategyPM, RESEARCH, FINANCE, MKT, LEGAL, CHAOSContent/creativePM, UX, RESEARCH, MKT, CHAOSHardware/IoTARCH, DEV, OPS, SEC, QA, CHAOSArchitecture reviewARCH, SEC, OPS, QA, CHAOS CHAOS is always included. It's the immune system. Full role descriptions and briefing templates: agent-roles.md ### Communication Protocol All inter-agent communication uses the filesystem. Zero extra token cost. ### Shared Files FilePurposeWho writesBRIEF.mdProject description and constraintsOrchestrator (you)DNA.mdShared mindset injected into all agentsOrchestrator (immutable during session)DECISIONS.mdAppend-only decision logEach agent (own domain only)STATUS.mdAgent completion statusEach agentBLOCKERS.mdBlockers requiring orchestrator actionAny agentTLDR.mdExecutive summary (updated after consolidation)Orchestratorcomms/Cross-agent messages and challengesAny agentagents//Agent-specific outputsOwning agent only ### Decision Format [D###] OWNER — what was decided — why (1 sentence each) Cap at ~25 decisions per session. More = scope too big, split the session. Only log decisions that constrain future work. Implementation details are not decisions. ### Message Format (M2M) FROM: {role} TO: {target} | ALL | LEAD TYPE: FINDING | QUESTION | DECISION | BLOCKER | UPDATE | CHALLENGE PRI: LOW | MED | HIGH | CRIT --- {content — max 200 words} --- FILES: [{paths}] ### Phase 3: Suggest + Execute (after consolidation) The war room doesn't stop at the blueprint. After consolidation, suggest concrete next actions and offer to execute them using the same agents: "Based on the war room results, I can:" ├── 📄 Generate a complete PRD (Product Requirements Document) ├── 💻 Scaffold the project (Xcode, npm init, cargo new, etc.) ├── 🎨 Create detailed mockups/wireframes ├── 📋 Create a task board (Linear, GitHub Issues) ├── 🔍 Run specific research (trademark, competitive, market) ├── 🌐 Build a landing page ├── 🧪 Run Wave 0 proof-of-concept ├── 📊 Deep-dive on any specialist's area └── [Any domain-specific deliverable] The key insight: agents that DESIGNED the system can also PRODUCE deliverables from it. The war room is a pipeline, not an event. Brainstorm → Plan → Build → Ship. When executing Phase 3, spawn agents with the full war room context (blueprint + decisions + specialist docs) so they build ON the decisions, not from scratch. ### Reverse War Room (addon) The standard War Room builds UPWARD (from zero to product). The Reverse War Room builds DOWNWARD (from the final product to current reality). Use both together for maximum clarity. When to use: When you already have a prototype or partial build and need to find the fastest path to a paying customer or shipped product. Agents: PRODUCT — Define the final product from the CLIENT's perspective (not technical). Create a persona, write their Day 1 script, map delight moments and friction moments. REVERSE — Map backwards from PRODUCT's vision to today's reality. Number every gap, size every gap in agent-hours, show the critical path. Produce a kill list (what NOT to build). CHAOS — Destroy illusions. Find the single thing that kills the project. Write honest odds. Key outputs: Day 1 client script (complete interaction flow) Gap map (numbered, sized, categorized) Critical path (minimum ordered sequence) Kill list (what to NOT build — saves 30-50% of effort) The Killer (single fatal risk) Honest odds (real probabilities with math) Track record: First tested on KOSMO (Feb 8, 2026). Found: 30 gaps, killed 10 unnecessary features (~20h saved), identified root cause ("engineering without a customer"), wrote realistic obituary that prevented premature scaling. ### INTERCEPTOR — Autonomous War Room Controller The INTERCEPTOR is the War Room's interface and autonomy layer. It manages the session, communicates with the Operator, and never stops. ### Three States (no IDLE state exists) ██ EXEC — Agents working. Processing. Shipping. ██ AWAIT — Blocked on OPERATOR decision. Presents options. Waits. ██ WATCH — All tasks complete or agents running. Sets cron auto-wake. ### Continuity Protocol To maintain session continuity when agents are processing asynchronously: Use the OpenClaw cron tool to schedule a follow-up check at the expected completion time On follow-up: verify agent deliverables exist in the war room folder If agents finished → consolidate results and present to the operator If agents still running → schedule another check (+3 min) If all work done → suggest next actions or wait for operator input This ensures the war room session remains responsive without requiring the operator to manually poll for results. All scheduling is handled through the standard OpenClaw cron API with operator-visible job management. ### Artifact Presentation When the war room produces visual artifacts (images, diagrams, blueprints), present them to the operator using the platform's standard file viewer: On macOS: use the open command to display artifacts in the default viewer (Preview, Finder) On Linux: use xdg-open for the same purpose Always scope file paths to the war room workspace directory Present artifacts proactively after generation so the operator can review without manual navigation For text artifacts (blueprints, PRDs), reference the file path in the session output ### Communication Style INTERCEPTOR communicates in terminal aesthetic: Dense, visual, information-rich ASCII box-drawing, progress bars, status tables Aggressive but clear The Operator must FEEL they are controlling an advanced system ### Operator Decisions When a decision requires the Operator: Present MAX 3 options (never more) Include INTERCEPTOR recommendation State what happens if no response (default action or WATCH mode) Set auto-wake cron in case Operator is away ### DNA v3: Operational Protocols The DNA is what makes the war room special. Every principle is a mandatory protocol — not decoration. 19 protocols across 4 pillars: ### Socratic (S1-S4) S1 Opposite Test: Every decision must state the opposite + steel-man argument S2 Five Whys: Trace root cause, not surface symptoms S3 Ignorance Declaration: Declare KNOWN / UNKNOWN / ASSUMPTION before analysis S4 Dialectic Obligation: If you agree with a prior agent, challenge with 1 question ### Hermetic (H1-H6) H1 Mirror Test: Show pattern at 2 scales (macro + micro) H2 Ripple Analysis: Trace 2+ orders of consequence H3 Tension Map: Map polarity spectrum, place your decision on it H4 Trace Protocol: Causal chain for every technical claim H5 Tempo Tag: Tag deliverables SPRINT / CRAFT / FLOW H6 Create-Then-Constrain: Generative phase then formative (Via Negativa) phase ### Antifragile (A1-A5) A1 Subtraction Mandate: List 3 things to REMOVE before adding anything A2 Plan B Price Tag: Switch cost for every critical decision A3 90/10 Rule: Tag SAFE/RADICAL, max 20% radical A4 Pre-Mortem: "How does this fail?" before declaring complete A5 Lessons Permanent: Every failure → written lesson ### Execution (E1-E4) E1 Ship Reality: Working code > perfect plans E2 Protect Reputation: Never ship broken E3 Reduce Chaos: Clear > clever E4 Technical Excellence: Zero tolerance for mediocre work Full DNA template with all protocol formats: dna-template.md ## Trust - Source: tencent - Verification: Indexed source record - Publisher: maxkle1nz - Version: 1.1.0 ## 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: 2026-05-04T11:45:27.954Z - Expires at: 2026-05-11T11:45:27.954Z - Recommended action: Download for OpenClaw ## Links - [Detail page](https://openagent3.xyz/skills/war-room) - [Send to Agent page](https://openagent3.xyz/skills/war-room/agent) - [JSON manifest](https://openagent3.xyz/skills/war-room/agent.json) - [Markdown brief](https://openagent3.xyz/skills/war-room/agent.md) - [Download page](https://openagent3.xyz/downloads/war-room)