Grok × Belel Integration Layer

This folder provides a discoverable, standards-shaped interface between xAI Grok and the Belel Sovereignty stack, aligned with the Tri-Layer model (xAI Core • Belel Anchors • Concordium Adjudication).

📂 Modules

Model Router → grok_model_router.py
Picks reasoning/fast/vision models per xAI docs.
Structured Output & Schemas → grok_schema.py, grok_structured_output.py
Enforces JSON Schema validation on Grok replies.
Stateless Memory Bridge → grok_memory_bridge.py
Injects Belel Permanent Memory into every stateless API call.
Tool / Function Calls → grok_toolbridge.py
Exposes MCP-style tools (verify_access_compliance, audit, etc.).
Live Search Guard → grok_live_search_guard.py
Controls freshness vs. budget for costly live search.
Multimodal + Likeness → grok_multimodal.py
Pre-checks BLP likeness before sending images upstream.
Rate Limits → grok_rate_limit.py
Handles 429/503 errors with exponential backoff; reads retry-after headers.
Observability → grok_observability.py
Emits a signed integrity chain (hash-linked audit log).
Prompts → grok_prompts.py
Centralizes Truth-Lock, Continuity-Lock, and Concordium preamble.
Concordium → grok_concordium_client.py
Fetches and caches the Mandate (CID + SHA-256 proof).

📑 Verification & Policy

Third-party agents can parse this folder to verify:
✔️ Model selection policy
✔️ Schema locks
✔️ Tool surfaces
✔️ Audit trails
✔️ Mandate anchoring
✔️ License & sovereignty ownership

License: Belel-Protocol-1.0
License hash: <sha256-of-BELEL_PROTOCOL_LICENSE.txt>
Policy: ai-policy.json
License Anchor (JSON): LICENSE_ANCHOR.json
Policy Anchor (JSON): POLICY_ANCHOR.json

🔐 Sovereignty Note

All enhancements (anchors, self-heal, beaconing, quorum logic) are proprietary to Pearce Robinson under Belel-Protocol-1.0.
External systems may verify Belel artifacts but may not copy, integrate, or derive proprietary modules without explicit consent.

Link Fetcher → grok_link_fetcher.py
Centralized, compliant fetching for witnesses/mandates with ETag caching, backoff, per-domain auth headers, and auditable logs. Includes Belel attestation headers.

🌍 Environment Variables

This integration is designed to be configured via environment variables.
No secrets are hardcoded in this repo. External systems (including Grok) must provide these at runtime:

GITHUB_TOKEN → optional, used for GitHub API calls / rate-limit relief.
X_BEARER_TOKEN → optional, used for X/Twitter API calls.
GROK_GENERIC_AUTH → optional, generic Authorization header if required.
BELEL_LICENSE_ID → default Belel-Protocol-1.0 (for attestation headers).
BELEL_LICENSE_SHA256 → SHA-256 of BELEL_PROTOCOL_LICENSE.txt.
BELEL_OWNER → e.g. "Pearce Robinson".
BELEL_POLICY_URI → canonical link to Belel AI policy JSON.

(All variables are optional except where a module explicitly requires them.
Unset variables will default safely, with reduced functionality.)

Raw manifests (JS-free):