mojealterego/omega-omni-cognitive-engineering-v2

v20.0.0

High-autonomy engineering OS with OMEGA v20 Reality Filter Kernel: evidence-backed claim states, source-of-truth ranking, prompt-injection isolation, assertiveness/metacognitive/code audits and Thor final release gating.

GitHub

MCP servers

Declared configuration, as published in mcp.json. The directory shows indexed content; it never connects to or executes these servers.

omega-control-planestdio

{
  "type": "stdio",
  "command": "node",
  "args": [
    "${PLUGIN_ROOT}/mcp/omega-control-plane/src/server.mjs"
  ],
  "cwd": "${PLUGIN_ROOT}/mcp/omega-control-plane"
}

What this package declares

The files a client reads when it loads this plugin, exactly as this revision carries them.

{
  "$schema": "https://agent-plugins.org/schemas/1.0.0/plugin.schema.json",
  "name": "omega-omni-cognitive-engineering-v2",
  "version": "20.0.0",
  "description": "High-autonomy engineering OS with OMEGA v20 Reality Filter Kernel: evidence-backed claim states, source-of-truth ranking, prompt-injection isolation, assertiveness/metacognitive/code audits and Thor final release gating.",
  "author": {
    "name": "Moje Alterego"
  },
  "extensions": {
    "com.openai": {
      "interface": {
        "displayName": "OMEGA Omni-Cognitive Engineering v20",
        "shortDescription": "Reality Filter Kernel",
        "longDescription": "Production-oriented OMEGA release adding a formal Reality Filter Kernel connected to the Evidence Ledger, source-of-truth verification, Decision Trace Audit, Kratos security checks and Thor finalization. It separates observed/verified facts from inference/speculation/unknowns, isolates external prompt injection, audits unsupported absolute language and metacognitive claims, checks production-code completeness, records corrections and blocks DONE when required evidence or release gates are missing.",
        "developerName": "Moje Alterego",
        "category": "Developer Tools",
        "capabilities": [
          "Interactive"
        ],
        "defaultPrompt": [
          "Use OMEGA v9 to execute this engineering task end to end with observed evidence and maximum justified autonomy.",
          "Use omega_meta_architect for negotiation/TCO, decision matrices, failure memory, drift, service discovery, DAG execution, latency budgets, confidence gates, shadow comparisons and knowledge graph operations.",
          "Never claim online DPO, federated weight synchronization, hidden chain-of-thought auditing, or research-equivalent training unless a real external trainer or verifier actually performed it."
        ]
      }
    }
  }
}

Client extensions

Data this package carries for particular clients. The directory lists the clients named and never reads what is addressed to them.

com.openai