Egeria Mcp

API | MCP Server

Apache Egeria open-metadata API + MCP Server for the Arbora ecosystem.

Version: 2.3.3

Documentation — Installation, deployment, usage across the API, CLI, and MCP interfaces, and guidance for provisioning the Apache Egeria platform are maintained in the official documentation.

Egeria is federated as the metadata / governance / lineage system-of-record alongside the epistemic-graph Knowledge Graph (the cognition/orchestration plane). This package provides the typed, deterministic Egeria tools a policy router calls, the bottom-up harvest that populates Egeria from the data estate, and the governed_route decision that is the federation delivering value.

Two hard invariants: the KG never becomes the lineage store; Egeria never orchestrates.

What it provides

• EgeriaApi (egeria_mcp.api.api_client_egeria) — a tolerant raw-httpx REST facade over the Egeria View Server (OMVS). No pyegeria runtime dependency (pyegeria's sync wrappers call asyncio.get_event_loop(), which raises on Python 3.14; httpx works on 3.11 and 3.14). It is the injected config["client"] for the KG egeria enrichment extractor and backs all MCP tools. Every call degrades to [] / a clear error rather than raising.
• 21 MCP tools (egeria-mcp console script): granular reads, broad action-dispatch readers across 11 OMVS services, governed_route, the data-store harvest, and write-gated create/classify/lineage tools. See docs/overview.md for the full list and the verified Egeria 6.0 REST contract.

Available MCP Tools

Auto-generated — do not edit (synced by the mcp-readme-table pre-commit hook).

Intent tools

41 intent tool(s) · 0 operation(s). Call an operation as act(action='<operation>', params={...}) (ask for reads); find(action='describe', params={'action': '<operation>'}) returns its arguments. An operation is enabled unless its <DOMAIN>TOOL toggle is set false. Auto-generated — do not edit.

Configuration (environment)

The table above is the required set. The bottom-up harvest connectors (ServiceNow, ERPNext, GitLab, Camunda, Keycloak, Grafana, Portainer, …) each read their own optional credential vars — every variable, grouped by source system with required-vs-optional separation, is documented in .env.example. Copy it to .env and populate only the connectors the operator use; blank connector credentials leave the corresponding harvest inactive.

Install & run

Pick the extra that matches what the operator want to run:

# Connector-focused MCP server (includes the shared graph engine)
uv pip install "egeria-mcp[mcp]"          # or: pip install -e ".[mcp]"

# Everything (development)
uv pip install "egeria-mcp[all]"          # or: python -m pip install "egeria-mcp[all]"

egeria-mcp                       # stdio MCP server (default transport)
egeria-mcp --transport http --host 0.0.0.0 --port 8000

The multi-stage docker/Dockerfile builds one immutable image (registry.example.invalid/egeria-mcp@sha256:<digest>) that installs egeria-mcp[all] and runs the egeria-mcp console script; docker/mcp.compose.yml runs it as a streamable-http service.

Run the bottom-up data-store harvest (needs write enabled):

EGERIA_PLATFORM_URL=https://your-egeria-platform:9443 EGERIA_ENABLE_WRITE=true \
  python -m egeria_mcp.harvest

Knowledge-graph database (epistemic-graph)

Both [mcp] and [agent] carry epistemic-graph[full] through the required Agent Utilities core. The [mcp] surface is connector-focused; [agent] also enables model orchestration.

Egeria is federated alongside the epistemic-graph Knowledge Graph: Egeria is the metadata / governance / lineage system-of-record, the KG is the cognition / orchestration plane (the KG never becomes the lineage store; Egeria never orchestrates). For production — or to share one knowledge graph across multiple agents — run epistemic-graph as its own database container. Deployment recipes (single-node + Raft HA), connection config, and the full database architecture (with diagrams) are documented in the epistemic-graph deployment guide.

MCP config

Register in the multiplexer under nickname eg (tools surface as eg__lineage, eg__governed_route, eg__catalog, …). See egeria_mcp/mcp_config.json.

Additional Deployment Options

egeria-mcp can run as a local stdio process or container, or behind a remote network boundary. The Deployment guide carries the detailed transport contract.

• Local container — launch a reviewed immutable image as a least-privilege stdio child with no listener or published port.
• Remote URL — connect through an operator-supplied authenticated HTTPS ingress. Keep its URL, outbound identity references, trust profile, and exact MCP_ALLOWED_HOSTS in AgentConfig.

Documentation

The complete documentation is published as the official documentation site and is the recommended reference for installation, deployment, and day-to-day operation.

AGENTS.md is the canonical contributor/agent guidance.

Deploy with agent-connector-sdk-deployment

Provision this package with the consolidated agent-connector-sdk-deployment workflow. It selects an installed-package, editable-source, or immutable-container path; records only runtime secret and TLS-profile references in AgentConfig; and runs doctor, registration, policy, observability, and rollback gates. Ask your agent to "deploy egeria-mcp with agent-connector-sdk-deployment".

The repository embeds no deployment profile, credential value, certificate path, or environment-specific endpoint. Supply those at runtime through AgentConfig and the configured secret provider.

Environment Variables

Package environment variables

Inherited agent-connector-sdk variables (apply to every connector)

136 package + 17 inherited variable(s). Auto-generated from .env.example + the shared agent-connector-sdk set — do not edit.

Governed capability contract

This package ships a compact canonical skill surface with specialist procedures kept as referenced workflows. The current MCP tools, skill metadata, connector_manifest.yml, ontology, mappings, shapes, fixtures, migrations, tool-schema fingerprints, and certification metadata form one versioned capability contract. Validate them together; do not rely on stale tool names or historical per-task skill wrappers.

Runtime endpoints, credentials, certificate trust, tenant identity, retention, and observability policy are deployment inputs and are never packaged values. See Configuration, trust, and privacy before enabling a network transport, connector ingestion, Arbora delegation, or trace export.