🐾 OpenClaw — Full System Report

1. What We Have (Current State)

What Works Right Now

2. Four-Layer Architecture

Signal Flow (gateway-first, no CLI nesting):

  Discord message → runtime.ts
                      ↓
              thetasteer-categorize (Ollama)
                      ↓
              Hardness 1-5 estimation (Ollama)
                      ↓
              Full 12×12 notation: 📡 B3 Tactics.Signal : 🔌 C1 Operations.Grid
                      ↓
              ┌───────┼───────┐
              │       │       │
           H1-2    H3-4     H5
           Ollama  Sonnet   Opus
              │       │       │
              └───────┼───────┘
                      ↓
              Gateway API (POST /v1/chat/completions)
                      ↓
              Response → Discord channel + spec update
  

3. Nine Cognitive Rooms

Terminal IPC Methods

Room Handoff Patterns

4. Skills Inventory (54 Total)

Workspace Skills (6 Custom)

Notable Bundled Skills (48)

Skills Needing Only an Install

5. TypeScript Modules (8 Source Files)

Skills (6 TypeScript Files)

6. Signal Flow & Hardness Routing

Hardness Routing Table

Structured Context Header

═══ TASK CONTEXT ═══
INTERSECTION: 📡 B3 Tactics.Signal : 🔌 C1 Operations.Grid
QUESTION: What does Signal mean in Grid?
HARDNESS: 4/5 → sonnet
ROOM: navigator
SPEC: [current spec state, first 500 chars]
═══ ROLLING CONTEXT ═══
[last 50 lines from room's Discord channel]
═══ PROMPT ═══
[actual user prompt]
  

Full 12x12 Axis Notation

7. Usability Stories (High Impact)

8. Best-Case Scenario: What We Could Build

Scenario A: Omni-Channel CEO

Scenario B: Visual Intelligence

Scenario C: Full Skill Ecosystem (54/54)

Scenario D: Mobile Command Center

Scenario E: FIM-Secured Autonomous Agent

Scenario F: Overnight Intelligence Machine

9. Known Issues & Fixes

10. TypeScript Documentation (Corrected)

tsconfig.json

{
  "compilerOptions": {
    "target": "ES2022",
    "module": "ESNext",
    "moduleResolution": "bundler",
    "strict": true,
    "esModuleInterop": true,
    "skipLibCheck": true,
    "forceConsistentCasingInFileNames": true,
    "outDir": "./dist",
    "declaration": true,
    "resolveJsonModule": true,
    "isolatedModules": true
  },
  "include": ["src/**/*.ts", "skills/**/*.ts"]
}

package.json

{
  "name": "@thetadriven/openclaw",
  "version": "0.1.0",
  "private": true,
  "type": "module",
  "engines": { "node": ">=22.0.0" },
  "scripts": {
    "start": "npx tsx src/runtime.ts",
    "doctor": "npx tsx src/doctor.ts",
    "dev": "npx tsx watch src/runtime.ts",
    "build": "tsc --noEmit"
  },
  "dependencies": {
    "discord.js": "^14.14.1",
    "ffmpeg-static": "^5.3.0"
  }
}

Core Types (src/types.ts)

// The contract every skill must implement
interface AgentSkill {
  name: string;
  description: string;
  initialize(ctx: SkillContext): Promise<void>;
  execute(command: unknown, ctx: SkillContext): Promise<SkillResult>;
}

// The context passed to every skill method
interface SkillContext {
  config: ConfigReader;           // Read openclaw.json values
  log: Logger;                    // Structured logging: info, warn, error
  fs: FileSystem;                 // read(path), write(path, content)
  shell: ShellExecutor;           // exec(cmd) → { stdout, stderr, code }
  callSkill(name: string, payload: unknown): Promise<SkillResult>;
}

// Every skill execute() returns this
interface SkillResult {
  success: boolean;
  message: string;
  data?: unknown;
}

// Terminal configuration for each cognitive room
interface TerminalEntry {
  room: string;          // "builder", "operator", etc.
  emoji: string;         // "🔨", "🎩", etc.
  app: string;           // "iTerm", "kitty", etc.
  processName: string;   // macOS process name for AppleScript
  ipc: IpcMethod;        // "iterm" | "kitty" | "wezterm" | "terminal" | "system-events"
  windowHint: string;    // Window/tab title search hint
}

// IPC dispatch methods
type IpcMethod = "iterm" | "kitty" | "wezterm" | "terminal" | "system-events";

// Orchestrator configuration from openclaw.json
interface OrchestratorConfig {
  channelCategory: string;     // "Cognitive Rooms"
  pollIntervalMs: number;      // 2000 (output polling frequency)
  taskTimeoutMs: number;       // 120000 (2 min max per task)
  stabilizationMs: number;     // 5000 (output stable = done)
}

// Task lifecycle
type TaskStatus = "pending" | "running" | "capturing" | "complete" | "failed" | "killed";
  

CategorizationResult (thetasteer-categorize.ts)

interface CategorizationResult {
  row: string;               // "B3"
  col: string;               // "C1"
  tile_id: string;           // "B3:C1" (compact, for compat)
  full_notation: string;     // "📡 B3 Tactics.Signal : 🔌 C1 Operations.Grid"
  semantic_question: string; // "What does Signal mean in Grid?"
  confidence: number;        // 0.0 - 1.0
  tier: "GREEN"|"RED"|"BLUE"; // GREEN=0.7+, RED=0.3-0.7, BLUE=<0.3
  hardness: number;          // 1-5
  target_model: "ollama"|"sonnet"|"opus";
  reasoning: string;         // Why this hardness level
}
  

Gateway API Reference

// Authentication
const GATEWAY_URL = "http://127.0.0.1:18789";
const AUTH_HEADER = "Bearer d9b07f435991e525a8c1c2d53c501a1f3df7863b4931c2b5";

// Chat Completion (OpenAI-compatible)
POST /v1/chat/completions
{
  "model": "anthropic/claude-sonnet-4-5",  // or "anthropic/claude-opus-4-6"
  "messages": [{ "role": "user", "content": "prompt" }]
}
→ { "choices": [{ "message": { "content": "response" } }] }

// WebSocket RPC (ws://127.0.0.1:18789)
{"method": "agents.create", "params": {"prompt": "...", "model": "..."}}
{"method": "agent.wait", "params": {"agentId": "..."}}
{"method": "agents.list"}
{"method": "agents.abort", "params": {"agentId": "..."}}
{"method": "models.list"}
{"method": "cron.add", "params": {"schedule": "0 8 * * *", "prompt": "..."}}
{"method": "config.get", "params": {"path": "agents.defaults.model"}}
{"method": "config.patch", "params": {"patch": {...}}}
  

SKILL.md Format

---
name: my-skill
description: What this skill does
metadata: {
  "openclaw": {
    "emoji": "🎯",
    "requires": {
      "bins": ["required-binary"],     // Must be on PATH
      "env": ["REQUIRED_API_KEY"],     // Must be set
      "config": ["some.config.path"]   // Must be truthy in config
    },
    "primaryEnv": "REQUIRED_API_KEY",
    "install": [
      { "kind": "brew", "formula": "package-name", "bins": ["binary-name"] }
    ]
  }
}
---

# My Skill

Instructions for the AI agent when this skill is loaded.
Use {baseDir} to reference the skill's directory.

## Scripts
- `scripts/my-script.sh` — Does the thing
  

Skill Locations (Precedence Order)

1. <workspace>/skills/            # Highest priority (our custom skills)
   ~/.openclaw/workspace/skills/
   ├── claude-flow-bridge/SKILL.md
   ├── output-capture/SKILL.md
   ├── system-control/SKILL.md
   ├── terminal-dispatch/SKILL.md
   ├── thetasteer-categorize/SKILL.md
   └── voice-memo-reactor/SKILL.md

2. ~/.openclaw/skills/              # Managed/installed skills
   (empty — use clawhub install)

3. node_modules/openclaw/skills/    # Bundled with npm package (48 skills)
   ├── github/SKILL.md
   ├── coding-agent/SKILL.md
   ├── tmux/SKILL.md
   └── ... (45 more)
  

11. Roadmap: Current → Dream State

Phase 1: Stabilize (This Week)

Phase 2: Expand (Next Week)

Phase 3: Integrate (Month 1)

Phase 4: Scale (Month 2+)