About

Imported from Rajbharti06/The-Orchestrator (skills/autonomous-loops/SKILL.md). Install upstream with npx skills add Rajbharti06/The-Orchestrator --skill autonomous-loops. Copyright stays with the author.

Autonomous Loops Skill

Name: autonomous-loops
Author: Rajbharti06

Orchestrator X can run self-improvement cycles without any human input.

How It Works

Every 10 minutes:
┌─────────────────────────────────────────────────────┐
│ 1. RUN EVAL SUITE (8 test cases against real LLM)  │
│    Planning, Coding, QA, Stack Detection tests     │
│                                                    │
│ 2. ANALYZE FAILURES                                │
│    Extract cause + fix for each failed case        │
│                                                    │
│ 3. WRITE LESSONS                                   │
│    Failed cases → lessonStore.addLesson()          │
│                                                    │
│ 4. UPDATE INSTINCT CONFIDENCE                      │
│    Weak areas → reduce confidence for related      │
│    instincts (Bayesian update)                     │
│                                                    │
│ 5. PRUNE EXPIRED PATTERNS                          │
│    Remove instincts < 0.2 confidence + 30 days old │
│                                                    │
│ 6. TRACK SCORE HISTORY                             │
│    Record score, trend, lessons added              │
└─────────────────────────────────────────────────────┘

Configuration

# Start with default interval (10 min)
POST /autonomous/start

# Custom interval (5 min)
POST /autonomous/start
{ "intervalMs": 300000 }

# Environment variable
ENABLE_AUTONOMOUS_LOOP=true npm start

Score Interpretation

Score	Interpretation	Action
80-100	Excellent	Maintain current patterns
65-79	Good	Minor improvements needed
50-64	Fair	Investigate weak areas
< 50	Poor	Review lessons and instincts

Trend Detection

After 3+ eval cycles, the system detects trends:

Improving (+3 points vs previous average) → current approach is working
Declining (-3 points vs previous average) → lessons may be stale or conflicting
Stable → system has reached equilibrium

Self-Improvement Mechanics

The system improves its memory (lessons + instincts), not its source code:

Safe: no risk of breaking changes
Measurable: track score history in dashboard
Reversible: lessons can be manually deleted
Transparent: all changes visible in /lessons and /instincts

Use it

Copy one of these into your project. Installing also returns the manifest and these snippets.

yaml

targets:
  - https://api.opensmartroute.ai/api/v1/registry/rajbharti06-the-orchestrator-autonomous-loops/manifest   # or paste the manifest below

Manifest

An Open Capability Manifest: the router reads it to know what this does, what it costs and when to pick it.

rajbharti06-the-orchestrator-autonomous-loops.ocm.jsonjson

{
  "ocm": "1",
  "id": "rajbharti06-the-orchestrator-autonomous-loops",
  "kind": "skill",
  "name": "autonomous-loops",
  "description": "Self-improvement loop patterns — eval → learn → improve cycles running without human input",
  "publisher": "Rajbharti06",
  "version": "1.0.0",
  "capabilities": {
    "domains": [
      "general"
    ],
    "tags": [
      "skill-md",
      "autonomous",
      "loop",
      "learning",
      "improvement",
      "github"
    ],
    "languages": [
      "en"
    ]
  },
  "quality_prior": 0.6,
  "examples": [
    "Self-improvement loop patterns — eval → learn → improve cycles running without human input"
  ],
  "primary": false,
  "metadata": {
    "source": {
      "provider": "github",
      "repository": "https://github.com/Rajbharti06/The-Orchestrator",
      "path": "skills/autonomous-loops/SKILL.md",
      "ref": "cab0d46afc03e2650d7a626a273641dd667977ce",
      "url": "https://github.com/Rajbharti06/The-Orchestrator/blob/cab0d46afc03e2650d7a626a273641dd667977ce/skills/autonomous-loops/SKILL.md",
      "key": "Rajbharti06/The-Orchestrator/skills/autonomous-loops/SKILL.md"
    }
  },
  "instructions": "# Autonomous Loops Skill\n\nOrchestrator X can run self-improvement cycles without any human input.\n\n## How It Works\n\n```\nEvery 10 minutes:\n┌─────────────────────────────────────────────────────┐\n│ 1. RUN EVAL SUITE (8 test cases against real LLM)  │\n│    Planning, Coding, QA, Stack Detection tests     │\n│                                                    │\n│ 2. ANALYZE FAILURES                                │\n│    Extract cause + fix for each failed case        │\n│                                                    │\n│ 3. WRITE LESSONS                                   │\n│    Failed cases → l",
  "cost": {
    "context_tokens": 585
  }
}

Fetch it by URL: GET /api/v1/registry/rajbharti06-the-orchestrator-autonomous-loops/manifest?version=1.0.0

Reviews

Star ratings from people who tried it. One review per account; edit yours any time.

No reviews yet. Install it, try it, and be the first to rate it.