Trust Center · Anti-drift transparency
Every capability count on every LYDOS marketing surface comes from one file: manifest/canonical-stats.json. This page renders that file verbatim so you can spot a mismatch before we do.
Version
v12.3.0-beta
Production status
controlled_beta
Manifest schema
v1.3 · last regen 2026-06-01
| Metric | Value | Source of truth |
|---|---|---|
| Agents | 120 | 13 Python-backed + 107 YAML-defined. 46 of these run as 24/7 autonomous engineers across 14 departments. Verifiable in: |
| Q-Engines | 245 | Unique Q-numbered engines registered in route handlers (Q1-Q250 with gaps). Verifiable in: |
| Sovereign Engines | 13 | The 13 isolation-first engines: DPI, LocalSec, DPI-H, CC, FIN, LSSA, KSL, TUP, GCE, ASR, DSL, LSIA, GCI. Verifiable in: |
| MCP tools | 162 | Tools registered with the Model Context Protocol server. Any MCP-compatible client (Claude Code, Cursor, Cline, VS Code Continue, etc.) sees these. Verifiable in: |
| AI CLIs | 6 | Shipped CLI bridge adapters: Claude Code, Cursor, Cline, OpenAI Codex CLI, Aider, VS Code Continue. Windsurf and GitHub Copilot are tracked as Roadmap — not claimed on any surface until adapters ship. Verifiable in: |
| LLM Providers | 12 | Backend providers the Q159 router dispatches to: Groq, Anthropic, OpenAI, Mistral, Gemini, DeepSeek, Z.AI, Qwen, Cohere, NVIDIA NIM, Kimi, Ollama. Per-user API keys live in the AES-256-GCM vault. Verifiable in: |
Two enforcement layers run on every commit:
core/integrations/canonical_stats.py asserts at load time that capability_counts.llm_providers equals len(SUPPORTED_LLM_PROVIDERS) in the live LLM registry. If a future edit moves one without the other, the backend fails to import.tests/test_sprint37_phase2_canonical_stats_consumer.py scans the marketing routes for hard-coded counts. Any new hard-coded number that should reference getPublicStatRows() fails CI.