ODGS governance enforcement — validation, regulation compilation, drift detection, signed S-Certs.
Deterministic governance enforcement at the AI agent boundary.
For engineers: See Quick Start below, or the guided walkthrough at metricprovenance.com/start.
For compliance and risk officers: The ODGS engine generates cryptographic audit trails (S-Certs).
For architectural clearance and certified packs: metricprovenance.com/pricing
Eight DAMA DMBOK pillars scored in about ten minutes — real MCP server, real maturity engine, click to watch on GitHub.
[!IMPORTANT] ODGS MCP Server v0.3.4 — Maturity Diagnostics + Certified Pack Access
governance_scorenow delegates to theodgs-maturity8-pillar DAMA DMBOK engine. Regulatory compilation, drift detection, and catalog synchronisation require a Certified Pack licence. The community tier — validation, scoring, conformance checking — remains open with no registration.
| Change | Detail |
|---|---|
governance_score | Delegates to odgs-maturity (8-pillar DAMA DMBOK) when installed; falls back to built-in heuristic. Returns 0–100 score with per-pillar gap analysis. |
| AuthGate | Community / Professional / Enterprise access via API key validation. 24h disk cache; workspace.yaml fallback for air-gapped deployments. |
| Licensed tools | compile_regulation, check_drift, detect_conflicts, narrate_audit, discover_bindings (Professional); harvest_sovereign_rules (Enterprise). |
The European Data Governance Maturity Benchmark 2026 recorded an average maturity of 37.6% across 99 enterprises — a 62.4% gap against current regulatory expectation.
governance_scoreapplies the same assessment methodology to your project.
The ODGS MCP Server connects any MCP-compatible AI agent to the ODGS Sovereign Validation Engine. The engine evaluates data operations against governance rules at runtime and produces S-Certs (Sovereign Certificates) — cryptographically signed, machine-verifiable records that a governance rule was applied.
This is the bridge between probabilistic AI inference and deterministic governance enforcement.
Organisations operating High-Risk AI Systems under EU AI Act Articles 10 and 12 require demonstrable, auditable data governance at the pipeline level. S-Certs provide that audit trail in a format suitable for regulatory submission.
Certified Sovereign Packs and the S-Cert Registry are available through Metric Provenance certified implementation partners. For architectural assessment: metricprovenance.com/pricing.
# Core validation capabilities (free, no account needed)
pip install odgs-mcp-server
# With maturity scoring
pip install odgs-mcp-server odgs-maturity
# Complete installation with LLM bridge capabilities
pip install "odgs-mcp-server[llm]"
The server operates over standard stdio transport, making it instantly compatible with any MCP client.
Add to claude_desktop_config.json:
{
"mcpServers": {
"odgs-governance": {
"command": "odgs-mcp-server",
"args": ["--transport", "stdio"],
"env": {
"ODGS_PROJECT_ROOT": "/path/to/your/odgs/project"
}
}
}
}
Add to .cursor/mcp.json:
{
"mcpServers": {
"odgs-governance": {
"command": "odgs-mcp-server",
"args": ["--transport", "stdio"],
"env": {
"ODGS_PROJECT_ROOT": "/path/to/your/odgs/project"
}
}
}
}
Regulatory compilation, certified packs, and catalog synchronisation require a Certified Pack licence. Provide the issued API key in your client configuration:
"env": {
"ODGS_API_KEY": "sk-odgs-...",
"ODGS_PROJECT_ROOT": "/path/to/your/odgs/project"
}
Licences are self-serve — buy at metricprovenance.com/pricing and your key is emailed on purchase:
| Tier | Price | Includes |
|---|---|---|
| Community | Free | validate_payload, governance_score, list_packs, conformance_check — no registration |
| Team | €990/yr | One certified regulation pack, signed + updated, plus all Professional tools |
| Professional | €2,490/yr | All certified packs, priority re-signing on regulatory change |
| Consultant | €4,990/yr | Professional + white-label / client-deliverable rights |
Enterprise deployments (Sovereign CA nodes, harvest_sovereign_rules) via certified partners.
| Tool | Description |
|---|---|
validate_payload | Validate data against ODGS governance rules, produce S-Cert |
validate_batch | Validate multiple payloads in one call |
list_packs | List available Certified Regulation Packs |
governance_score | Score governance maturity (0–100) across 8 DAMA pillars with gap analysis |
conformance_check | Run ODGS conformance self-check (L1/L2) |
| Tool | Description |
|---|---|
download_pack | Download and cache certified regulatory rule packs locally |
compile_regulation | Convert regulation text → validated ODGS rule JSON |
check_drift | Detect semantic drift in governance definitions |
detect_conflicts | Find contradictions between regulatory rules |
narrate_audit | Convert S-Cert → human-readable narrative |
discover_bindings | Auto-generate physical data mappings from catalogs |
| Tool | Description |
|---|---|
harvest_sovereign_rules | (Flint Bridge) Automatically extract and mint rules from data stores — preview, pending verification against the harvester interface |
The ODGS MCP Server is designed for zero-trust, local-first execution. All data validation happens strictly on your machine. No sensitive data leaves your perimeter.
flowchart TB
Agent[AI Agent\nClaude/Cursor/Custom]
subgraph "ODGS MCP Server"
Auth[AuthGate]
Val[OdgsInterceptor v6]
Maturity[odgs-maturity\n8-pillar DAMA]
LLM[OdgsLlmBridge]
Flint[Flint Bridge]
end
Reg[(S-Cert Registry\nregistry.metricprovenance.com)]
Project[(Local ODGS Project)]
Agent -- "JSON-RPC (stdio)" --> Auth
Auth -- "Validate Key (HTTPS, cached 24h)" --> Reg
Auth --> Val
Auth --> Maturity
Auth --> LLM
Auth --> Flint
Val -- "Reads Rules" --> Project
Val -- "Generates" --> SCert[S-Cert]
Maturity -- "Scores pillars" --> Project
LLM -- "Compiles Regulations" --> Project
Flint -- "Harvests Sovereign Rules" --> Project
Cryptographically signed rule bundles, each mapped to a specific regulatory instrument. Packs are compiled from authoritative legislative text and certified by Metric Provenance.
| Pack | Regulation | Status |
|---|---|---|
| EU AI Act | Regulation (EU) 2024/1689 | ✅ Certified |
| DORA | Digital Operational Resilience Act | ✅ Certified |
| GDPR | General Data Protection Regulation | ✅ Certified |
| CSRD | Corporate Sustainability Reporting Directive | ✅ Certified |
| NIS2 | Network and Information Security Directive | ✅ Certified |
| Basel III | Basel Committee on Banking Supervision | ✅ Certified |
Full catalogue of 15+ packs. Licensing and deployment via metricprovenance.com/pricing.
| Variable | Description | Default |
|---|---|---|
ODGS_PROJECT_ROOT | Path to ODGS governance definitions | Current directory |
ODGS_API_KEY | API key for Professional/Enterprise access | None (community) |
ODGS_REGISTRY_URL | Registry endpoint for key validation | https://registry.metricprovenance.com |
ODGS_CACHE_DIR | Local cache for downloaded packs | ~/.odgs/cache |
The Open Data Governance Standard (ODGS) is an open protocol for deterministic data governance enforcement. ODGS is on a path toward formal standardization.
pip install odgsApache 2.0 — see LICENSE.
The protocol engine and MCP server are open source. Certified Regulation Packs are issued under a commercial licence.
Source-derived launch command. Check the maintainer’s required arguments and credentials before running:
uvx odgs-mcp-serverMerge this template into ~/Library/Application Support/Claude/claude_desktop_config.json. Keep existing servers. Add any arguments, credentials, and permissions required by the maintainer; this template has not been install-tested.
{
"mcpServers": {
"com-metricprovenance-odgs-mcp-server": {
"command": "uvx",
"args": [
"odgs-mcp-server"
]
}
}
}Restart Claude Desktop completely for changes to take effect. Confirm the server appears connected in the client’s tool list, then try a read-only example from its documentation.
Claude Desktop setup referenceodgs-mcp-serverpypicom.metricprovenance/odgs-mcp-server works with any MCP-compatible client. Copy the config snippet from the Configuration section above and add it to the file shown for your client, then restart the application.
~/Library/Application Support/Claude/claude_desktop_config.jsonRestart Claude Desktop completely for changes to take effect.~/.cursor/mcp.jsonRestart Cursor for changes to take effect..vscode/mcp.jsonReload VS Code window for changes to take effect.~/.codeium/windsurf/mcp_config.jsonRestart Windsurf for changes to take effect..mcp.jsonSave at the project root, then start Claude Code in that project and review the MCP server approval prompt. Keep real credentials out of shared files.