cern-opendata-mcp-server

Get CERN Open Data Records

cern_opendata_get_records

Read-onlyIdempotent

Fetch full metadata for 1-20 records in one call, by recid, DOI, CMS dataset path (/Primary/Era/TIER) or documentation slug. Returns the description, run periods, collision and distribution details, related records, a software-environment summary, the license and a ready citation, plus what a record states of its variable dictionary, physics category, pile-up, keywords, and LHCb magnet polarity and stripping. Documentation and news pages include their markdown body in slices of at most 30,000 characters; body_offset with that one id reads on from where a slice stops. Each response holds to 64,000 bytes, except that the first record always comes back whole: records past the budget, and ids past the first 20, are listed under deferred, to pass back as ids. File lists are not included; use cern_opendata_list_files. Identifiers that do not resolve come back under missing with guidance; they do not fail the call.

Input Schema

TableJSON Schema

Name	Required	Description	Default
`ids`	Yes	Identifiers to resolve: up to 100, of which the first 20 distinct are looked up and the rest returned under deferred. Forms may be mixed; exact duplicates collapse.
`body_offset`	No	With exactly one documentation or news id: where its body slice starts, in characters (UTF-16 code units), usually the body_next_offset an earlier call returned. 0 (the default) is the body's start and is accepted with any ids; above 0 needs exactly one id.

Output Schema

TableJSON Schema

Name	Required	Description
`error`	No	Present when the call failed. Absent on success.
`notice`	No	Caveats about the returned records: a documentation body with more to read, records deferred by the response budget, or ids past the first 20 that were not looked up, each with the call that continues it.
`missing`	No	Identifiers that resolved to no record; empty when every id resolved.
`records`	No	Resolved records, in the order of their first matching input.
`deferred`	No	Inputs left for another call, in input order: those whose records were left out to hold the response to 64,000 bytes (never the first record, which is returned whole even when larger), then ids past the first 20, which were not looked up. Pass them as ids to fetch them. Empty when nothing was left out.

Schema Changelog

Changes observed during successful MCP inspections.

Changed4 schema fields changedOct 18, 2026

changedInput schema / properties / ids / description

Previous value: -"Identifiers to resolve, 1-20: an array, or one comma-separated string. Forms may be mixed; duplicates collapse."New value: +"Identifiers to resolve: up to 100, of which the first 20 distinct are looked up and the rest returned under deferred. Forms may be mixed; exact duplicates collapse."

changedInput schema / properties / ids / maxItems
```
Previous value: -20New value: +100
```

changedOutput schema / properties / deferred / description

Previous value: -"Inputs whose records were left out to hold the response to 64,000 bytes, in input order; pass them as ids to fetch those records. Empty when every resolved record is returned."New value: +"Inputs left for another call, in input order: those whose records were left out to hold the response to 64,000 bytes (never the first record, which is returned whole even when larger), then ids past the first 20, which were not looked up. Pass them as ids to fetch them. Empty when nothing was left out."

changedOutput schema / properties / notice / description

Previous value: -"Caveats about the returned records: a documentation body with more to read, or records deferred by the response budget, each with the call that continues it."New value: +"Caveats about the returned records: a documentation body with more to read, records deferred by the response budget, or ids past the first 20 that were not looked up, each with the call that continues it."

First observedOct 2, 2026

TDQS

A4.6/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations only declare readOnly/idempotent/openWorld; the description adds substantial non-obvious behavior: an effective 20-id lookup ceiling, 64,000-byte response budget, the guarantee that the first record is always returned whole, deferral of over-budget records/ids, non-resolving ids surfacing under 'missing' rather than failing, and 30,000-character body slicing. This is well beyond what the annotations convey and is consistent with them.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness4/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Purpose and identifier forms are front-loaded, and nearly every sentence carries substantive constraints. However, the long nested clause stringing together returned fields is dense and slightly harder to parse than it needs to be, keeping it short of a 5.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness5/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a complex batch-fetch tool with an output schema, the description still covers everything an agent needs to call it correctly — accepted id forms, batch ceiling, pagination via body_offset, response-size handling, and error semantics. Nothing material is left to inference.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so baseline is 3, but the description adds meaning the schema does not: the first-20-distinct lookup rule, duplicate collapsing, and the precise interaction of body_offset with a single documentation/news id. The only tension is '1-20 records in one call' versus the schema's maxItems of 100, which the deferred sentence partly reconciles.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource ('Fetch full metadata for ... records'), enumerates the four accepted identifier forms (recid, DOI, CMS dataset path, documentation slug), and explicitly contrasts itself with a sibling ('File lists are not included; use cern_opendata_list_files'). An agent can distinguish this from get/search siblings without opening the schema.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines4/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Gives explicit routing for one sibling ('use cern_opendata_list_files') and states the condition that governs body_offset ('with that one id reads on from where a slice stops'). It does not explicitly contrast against cern_opendata_search_records, so the when-not-this-tool story is only partial, but usage context is clear.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All cern-opendata-mcp-server tools Inspect

Try in Browser

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.