Entity Enricher MCP Server
A hosted, remote Model Context Protocol server for
Entity Enricher — structured knowledge extraction with multiple
LLM providers. Connect Claude Desktop, Claude Code, Cursor, claude.ai or any MCP-compatible
client and, from inside a chat:
- Author JSON schemas — generate a sample entity, turn it into a schema, refine it in natural language.
- Enrich entities — single or batch (up to 100), against your schemas, with any of your configured models.
- Enrich multilingually — schemas with localized text fields get per-language values in every language you request, in one pass.
- Fuse multi-model results — conflicts detected field-by-field, resolved by voting or LLM arbitration.
- Benchmark models on your own data — saved scenarios, gold references, auto-scored quality / cost / speed.
- Ground enrichments in documents — upload PDFs, images or audio and attach them to any flow.
No install, no local process: the server runs at https://entityenricher.ai/api/mcp/
(streamable HTTP). This repository holds the public documentation and ready-to-use client
configs; the server implementation lives in the Entity Enricher platform.
Listed on the official MCP Registry as
ai.entityenricher/enricher (see server.json).
Quickstart
Option 1 — OAuth (recommended)
For claude.ai, Claude Code, Cursor, and any MCP client that implements the standard OAuth
flow. No API key to create or paste — the client discovers the authorization server
automatically, your browser opens the Entity Enricher consent screen, and the connection acts
on your behalf with your own role. Revoke it anytime under Settings → API Keys → Connected
Apps.
Claude Code
claude mcp add --transport http entity-enricher https://entityenricher.ai/api/mcp/
Then run /mcp in a session and pick Authenticate — your browser opens the consent page.
More options (project .mcp.json, API-key fallback): examples/claude-code/
claude.ai
Settings → Connectors → Add custom connector with URL
https://entityenricher.ai/api/mcp/, then click Authorize on the consent screen.
Walkthrough: examples/claude-ai-remote.md
Cursor / other OAuth-capable clients
Register the URL with no headers and the client prompts you to sign in:
examples/cursor/mcp.json
Option 2 — API key (static JSON configuration)
For clients configured via a JSON file rather than an interactive sign-in (Claude Desktop,
Continue, Zed) — and for headless/CI use.
-
In the Entity Enricher web UI: Settings → API Keys → New
organization access key. Pick a role — operator (read-mostly), editor (create/edit
schemas), or owner (full control, required for benchmarks). Copy the ent_… value; it's
only shown once.
-
For Claude Desktop, edit ~/Library/Application Support/Claude/claude_desktop_config.json
(macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"entity-enricher": {
"url": "https://entityenricher.ai/api/mcp/",
"headers": { "X-API-Key": "ent_your_key_here" }
}
}
}
Restart Claude Desktop. Full file: examples/claude-desktop/
Try it
List my Entity Enricher schemas, then enrich "Sanofi" against the pharmaceutical company
schema in English and French.
Claude discovers the tools automatically, confirms the model and schema choice with you, and
returns the structured result inline.
Recipes
Copy-paste chat walkthroughs of the three flows people ask about most:
37 tools, spanning the full schema-authoring and enrichment surface:
| Category | Tool | Description |
|---|
| Discovery | list_models | List the LLM models, languages, strategies, and (when the org has a plan with limits) the operational profile_limits available to the caller. |
| Schemas | generate_sample | Generate a realistic sample entity JSON from an entity-type description — the entry point of the schema-authoring loop. |
| Schemas | list_schemas | List saved JSON schemas in your organization, pinned ones first. |
| Schemas | get_schema | Fetch the full content of a saved schema by ID, including all properties, key fields, expertise domains, and validation rules. |
| Schemas | create_schema_from_sample | Generate and auto-save a JSON schema whose paths and types strictly follow an approved sample. |
| Schemas | edit_schema | Modify an existing saved schema using a natural-language prompt, via an LLM (e.g. 'add a regulatory_status field marked as a key property', 'remove the founding_year field', 'mark… |
| Schemas | save_schema | Persist a schema you authored directly (no LLM call, no cost) as a new saved schema. |
| Schemas | update_schema | Update a saved schema without an LLM call: rename, replace the schema_content, change tags, pin/unpin, or toggle non-determinism analysis. |
| Schemas | delete_schema | Soft-delete a saved schema by ID (restorable server-side shortly after; permanent deletion stays in the web UI). |
| Schemas | analyze_sample_determinism | Flag properties in a sample entity whose enriched value would DIFFER across models or reruns (non-deterministic names: temporal / ambiguous / subjective / multi-valued). |
| Schemas | analyze_schema_determinism | Analyze a saved schema and write a non_determinism flag onto each property whose enriched value is likely to vary across models/runs (badge in the editor). |
| Enrichment & fusion | start_batch_enrichment | Start an asynchronous batch enrichment against a JSON schema and return {job_id, total} immediately. |
| Enrichment & fusion | fetch_entities | Fetch a JSON array of entities from an external REST API (GET), server-side — the input step before start_batch_enrichment. |
| Enrichment & fusion | enrich_entity | Run a multi-model enrichment of a single entity against a JSON schema, returning the fused/best structured result. |
| Enrichment & fusion | retry_expertises | Re-run only the FAILED expertise domains of an existing multi-expertise enrichment record, merging the recovered values back into the record — no re-payment for the domains that… |
| Enrichment & fusion | merge_records | Merge 2+ enrichment records of the same entity into one fused result — the manual / re-run counterpart of the automatic fusion that follows a 2+ model enrich_entity or batch run. |
| Job control | get_job_status | Poll the status of an asynchronous LLM job — the middle step of every start → poll → fetch flow (start_batch_enrichment, generate_sample, run_benchmark, retry_expertises). |
| Job control | cancel_job | Cancel a pending, running, or paused LLM job started by start_batch_enrichment, generate_sample, run_benchmark, or retry_expertises. |
| Job control | answer_job_question | Answer the clarification questions of a paused job and resume it — the reply half of the interactive loop used by generate_sample's document-grounded planner (get_job_status… |
| Records & stats | list_records | List past enrichment records in your organization, most recent first. |
| Records & stats | get_record | Fetch a single enrichment record by ID, including the full structured output, validation errors, prompts/responses, and metrics. |
| Records & stats | get_stats | Aggregated statistics over your organization's enrichment records: totals, success rate, token usage, and cost summary. |
| Benchmarks | list_benchmark_scenarios | List the organization's benchmark scenarios (saved, reusable enrichment tests: schema + entity + strategy + scoring config). |
| Benchmarks | get_benchmark_scenario | Fetch one benchmark scenario with its per-model results (quality / cost / speed scores; results whose config_hash differs from the scenario's are stale — re-run those models). |
| Benchmarks | create_benchmark_scenario | Create a benchmark scenario — a reusable model test. |
| Benchmarks | update_benchmark_scenario | Update a benchmark scenario. |
| Benchmarks | set_benchmark_reference | Save a scenario's gold reference — the expected output each model result is scored against, and the gate between create_benchmark_scenario and run_benchmark. |
| Benchmarks | delete_benchmark_scenario | Delete a benchmark scenario and its results. |
| Benchmarks | run_benchmark | Launch a benchmark run — the final step of the benchmark lifecycle: execute the scenario's task (enrichment / sample generation / schema generation) with each selected model… |
| Attachments | upload_attachment | Upload a file (base64-encoded) so it can be used as source material in LLM flows. |
| Attachments | delete_attachment | Permanently remove an attachment from the server by id. |
| Database Sync | list_database_syncs | List the database syncs registered on a saved schema, with pending delta counts. |
| Database Sync | create_database_sync | Connect a database to a saved schema — the opt-in that turns enrichments into relational SQL deltas the user applies to their own PostgreSQL/MySQL/SQLite with the ee-database CLI… |
| Database Sync | delete_database_sync | Delete a database sync and its queued deltas. |
| Database Sync | create_database_credential | (Re)issue the sync-client credential of a database sync — the pairing step of the ee-database CLI workflow. |
| Database Sync | fetch_database_deltas | Fetch the next FIFO window of SQL deltas for a database sync. |
| Database Sync | ack_database_deltas | Acknowledge applied database deltas up to an id: releases the lease and, per the database's options, purges delivered copies and fully-delivered entity state. |
Tool behaviour is identical to the REST endpoints they wrap — same validation, billing and plan
limits as the web app. Write tools require the editor role; benchmark tools require
owner plus a plan that includes Model Benchmarks.
Resources
Resources let the client browse data without a tool call — both render as Markdown.
| Resource | URI template |
|---|
| Saved schema | enricher://schemas/{schema_id} |
| Enrichment record | enricher://records/{record_id} |
The async job pattern
MCP tools can't stream, so long-running work is split into start → poll → fetch:
- A start tool (
start_batch_enrichment, generate_sample, run_benchmark,
retry_expertises) returns a job_id immediately.
get_job_status(job_id) polls progress; paused jobs carry clarification questions that
answer_job_question resolves; cancel_job aborts.- Persisted outputs are fetched with
list_records(job_id=…) (or the feature's own read
tool, e.g. get_benchmark_scenario).
Jobs are held in a bounded in-memory manager — an unknown job_id means the job finished long
ago; go straight to the records.
Interactive classification resume
The feature that only an interactive client unlocks. With a classification model enabled, a
pre-flight check verifies the entity matches the schema type. On a mismatch the tool returns a
non-error response instead of failing:
{
"success": false,
"error_code": "classification_warning",
"message": "Pre-flight classification rejected the entity. ...",
"classification": {
"status": "mismatch",
"reasoning": "Titan is a moon of Saturn, not a planet.",
"confidence": 0.97
},
"job_id": "..."
}
Claude surfaces the reasoning, asks you to confirm, and retries with
force_after_classification_warning=true. Workflow connectors (n8n, Make) have to auto-cancel
here — a chat can just ask.
Error codes
Errors are structured dicts with an error_code field the client can pattern-match on:
error_code | When |
|---|
invalid_request | Malformed UUID, mutually exclusive args, body validation failure. |
prompt_limit_reached | Daily/weekly/monthly prompt quota exhausted (HTTP 402), with period + usage details. |
insufficient_credits | Credit balance too low to start the job (HTTP 402), with balance + purchase URL. |
model_limit_exceeded / language_limit_exceeded | More models/languages requested than the plan allows (HTTP 402). |
concurrent_job_limit_reached | Too many active jobs for the org — wait or upgrade. |
classification_warning | ⚡ Non-error: pre-flight classifier rejected the entity (see above). |
benchmarks_not_in_plan | Benchmark tools need the owner role + a plan with Model Benchmarks (HTTP 403). |
enrichment_timeout / schema_generation_timeout | Job exceeded its timeout — try fewer models. |
schema_generation_failed | Upstream LLM error (HTTP 502). |
cancelled | Job cancelled mid-run (HTTP 499). |
not_found | Schema or record ID doesn't exist in your org. |
Authentication details
- OAuth 2.1 (recommended) — any MCP client implementing the standard auth spec (claude.ai,
Claude Code, Cursor, MCP Inspector) discovers it automatically: standard discovery via
/.well-known/oauth-protected-resource, dynamic client registration, PKCE, browser consent.
No key to create or paste. Tokens are audience-bound and instantly revocable under
Settings → API Keys → Connected Apps.
- X-API-Key — for clients configured via a static JSON file:
ent_… organization access
keys, created in Settings → API Keys. The role attached to the key (operator / editor /
owner) gates which tools succeed.
Links
About this repository
This repo contains the public documentation and client examples for the Entity Enricher MCP
server. The server itself is embedded in the Entity Enricher platform and maintained in the
main (private) monorepo; this repo is synced from it as a git subtree. Issues and discussions
are welcome here.
Licensed under the MIT License.