heretek-openclaw/skills/thought-loop

Thought Loop Skill

Structured thought generation from environmental deltas for the OpenClaw agent collective.

Overview

The Thought Loop skill enables continuous autonomous thinking by:

1. Detecting Changes - Monitoring file system, databases, external sources, and agent states
2. Generating Thoughts - Creating structured thoughts from detected changes
3. Scoring Confidence - Assigning confidence levels to each thought
4. Recommending Actions - Suggesting appropriate responses to thoughts

Installation

cd skills/thought-loop
npm install  # No external dependencies required

Usage

Full Thought Loop Cycle

# Run complete cycle (detect + generate)
node thought-loop.js run

# Output as JSON
node thought-loop.js run --json

Detect Changes Only

# Detect environmental changes
node thought-loop.js detect

# JSON output
node thought-loop.js detect --json

Generate Thoughts from Deltas

# Generate thoughts from provided deltas
node thought-loop.js generate --deltas '[{"type":"file_created","path":"./test.md"}]'

Generate Idle Thoughts

# Generate reflective thoughts when idle
node thought-loop.js idle --agent steward

Thought Types

Type	Description	Example Trigger
discovery	New file/resource detected	file_created
update	Existing resource modified	file_modified
alert	Resource deleted or agent offline	file_deleted
external_awareness	External event detected	external_cve
reflection	Internal self-reflection	idle
state_change	Database state change	db_modified

Example Output

{
  "deltas": [
    {
      "type": "file_created",
      "path": "/workspace/new-feature.md",
      "timestamp": "2026-03-31T00:00:00Z",
      "size": 1024
    }
  ],
  "thoughts": [
    {
      "id": "thought_1234567890_abc123",
      "type": "discovery",
      "trigger": "file_created",
      "subject": "new-feature.md",
      "observation": "New file created: /workspace/new-feature.md",
      "implication": "May affect active or pending proposals",
      "recommendation": "broadcast_thought",
      "confidence": 0.7,
      "timestamp": "2026-03-31T00:00:00Z",
      "agent": "steward",
      "metadata": {
        "path": "/workspace/new-feature.md",
        "size": 1024
      }
    }
  ],
  "timestamp": "2026-03-31T00:00:00Z",
  "agent": "steward"
}

Configuration

Set environment variables to customize behavior:

WORKSPACE_ROOT="/workspace"           # Root directory to monitor
CURIOSITY_DIR="/workspace/.curiosity" # Curiosity engine directory
DELTA_STATE_FILE="/tmp/delta-state.json" # State file location
AGENT_NAME="steward"                  # Current agent name
MAX_IDLE_THOUGHTS=3                   # Max idle thoughts to generate

Programmatic Usage

const { ThoughtLoop } = require('./thought-loop');

const loop = new ThoughtLoop('steward');

// Run full cycle
const result = await loop.run();
console.log(result.thoughts);

// Detect changes only
const deltas = await loop.detect();

// Generate thoughts from deltas
const thoughts = loop.generate(deltas);

// Generate idle thoughts
const idleThoughts = loop.idle();

Integration

The Thought Loop skill integrates with:

• Consciousness Module - Thoughts broadcast to global workspace
• Triad System - Triggers deliberation for high-importance thoughts
• Memory System - Thoughts stored in episodic memory
• Self-Model - Updates cognitive state based on thoughts

License

MIT