MCP server

Status

The MCP server is planned for Phase 5 of the CASRAI platform roadmap. The endpoint is not yet live. This page documents the intended shape so that prospective consumers — agent frameworks, IDE assistants, and editorial-workflow bots — can plan against a stable contract.

What MCP is

The Model Context Protocol is an open specification, originally published by Anthropic and now implemented by multiple agent runtimes, for connecting language-model agents to external tools and data sources. An MCP server advertises a small, typed set of tools and resources; an MCP client (an agent host) discovers those tools and invokes them on the model's behalf. See the Model Context Protocol specificationfor the wire-level details and the canonical client and server SDKs.

The protocol matters here because it is the first widely-adopted convention for letting an agent ask, in a single round-trip, "give me the authoritative definition of research integrityfrom the CASRAI Dictionary" — and receive back a typed, citation-bearing result rather than a best-effort hallucination from training data.

Why a dictionary belongs behind an MCP server

Controlled vocabularies are a classic retrieval-augmented generation target. They are small enough to enumerate, large enough to exceed a useful prompt budget, and they change. An MCP tool that hands an agent the canonical definition, examples, counter-examples, aliases, and relationships for a term — in JSON, with a stable identifier — gives the agent something it can cite. That is materially better than a paragraph of paraphrase, both for the human reading the agent's output and for downstream systems that may parse it.

The same argument applies to the CRediT contributor roles. An editorial assistant that drafts an "Author contributions" paragraph for a manuscript should be reading the 14 role definitions from the canonical source, not from a training snapshot that may predate ANSI/NISO Z39.104-2022 or its 2026 revision.

Planned transport

The CASRAI MCP server will run on the existing mcp-gateway Cloudflare Worker that already fronts the rest of the platform. The transport is the standard MCP-over-HTTP variant with Server-Sent Events for streamed responses. The worker handles per-key rate limiting, caching, and the translation from MCP tool calls into the underlying GraphQL endpoint and REST endpoints.

Planned endpoint

text

https://mcp.casrai.org/sse

Planned tools

The initial tool set is small on purpose. Each tool wraps a single, well-defined dictionary operation; the agent composes them as needed.

dictionary.lookup_term

json

{
  "name": "dictionary.lookup_term",
  "description": "Return the canonical definition, examples, counter-examples, aliases, and relationships for a CASRAI Dictionary term, identified by slug.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "slug": { "type": "string", "description": "Term slug, e.g. 'research-integrity'." }
    },
    "required": ["slug"]
  }
}

dictionary.search_terms

json

{
  "name": "dictionary.search_terms",
  "description": "Full-text search across dictionary terms, optionally constrained to a domain. Returns slugs and titles, not full bodies.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "query":  { "type": "string" },
      "domain": { "type": "string", "description": "Optional domain slug." },
      "limit":  { "type": "integer", "minimum": 1, "maximum": 50, "default": 10 }
    },
    "required": ["query"]
  }
}

credit.list_roles

json

{
  "name": "credit.list_roles",
  "description": "Return the 14 CRediT contributor roles with canonical NISO URIs and short definitions.",
  "inputSchema": { "type": "object", "properties": {} }
}

credit.role_definition

json

{
  "name": "credit.role_definition",
  "description": "Return the full definition for one CRediT role, by slug.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "slug": { "type": "string", "description": "Role slug, e.g. 'conceptualization'." }
    },
    "required": ["slug"]
  }
}

dictionary.object_template

json

{
  "name": "dictionary.object_template",
  "description": "Return an object template (composite record) and its field list, by slug.",
  "inputSchema": {
    "type": "object",
    "properties": {
      "slug": { "type": "string" }
    },
    "required": ["slug"]
  }
}

Planned resources

Alongside tools, the server will advertise a small set of MCP resources for whole-set browsing: casrai://dictionary/domains, casrai://credit/roles, and casrai://dictionary/release-notes/latest. Resources are listed by the client at connection time and read on demand; agents can include them in context without a tool call.

Authentication

Read-only tools will accept anonymous traffic, subject to the same per-IP rate limit as the GraphQL and REST surfaces. Higher quotas, and any future write tools, will require an MCP API key issued through the same Application Password mechanism documented on the GraphQL endpoint page.

Versioning

The tool surface follows the same versioning rules as the rest of the platform: additive changes ship without a version bump; breaking changes are announced on the API changelog at least one quarter before they take effect. The MCP server advertises its protocol version on the initialise handshake per the upstream MCP specification.