MCP server for executing code against multiple API backends (REST, GraphQL, SPARQL) with auto-staging, schema discovery, and sandboxed V8 isolate execution
This server exposes 13 tools with significant definition gaps. User-facing tools (search, execute variants) have reasonable descriptions (150-250 chars), but 7 internal proxy tools lack meaningful descriptions or are underdocumented. Most tools lack explicit input schema validation in the visible code, schemas are inferred from Zod definitions rather than explicitly registered. Parameter descriptions exist but are often generic ('JavaScript code to execute'). No output schemas are documented. Error handling is absent from tool definitions. The server prioritizes code-execution patterns over safe, constrained tool design.
Execute JavaScript code in a sandboxed V8 isolate with API proxy helpers (api.get/api.post), catalog search functions, and auto-staging for large responses. Code runs with access to searchSpec(), listCategories(), getEndpoint() helpers and the full API catalog or OpenAPI spec.
Execute JavaScript code in a sandboxed V8 isolate against a GraphQL endpoint. Code runs with gql.query(queryString, variables?), schema introspection helpers (schema.types(), schema.type(), schema.search()), and optional REST API proxy (api.get/api.post) for hybrid GraphQL+REST servers.
Execute JavaScript code in a sandboxed V8 isolate against a SPARQL endpoint. Code runs with sparql.query(queryString, opts?), sparql.ask(queryString), sparql.raw(queryString, opts?), and access to common ontology prefixes (rdfs, owl, xsd, skos, dcterms, void, obo, uberon, go, ncbigene, efo, obi, sio, up, ensembl, bgee).
Search the API catalog (catalog mode) or OpenAPI spec (OpenAPI mode). Returns matching endpoints with full parameter docs. Use this to discover API capabilities before calling <prefix>_execute.
Internal proxy handler for REST API calls (api.get, api.post) executed within the isolate. Routes HTTP requests through the host's apiFetch adapter with auto-staging for responses >30KB.
Internal proxy handler for GraphQL queries executed within the isolate. Routes GraphQL requests through the host's gqlFetch adapter with auto-staging for large responses.
Internal proxy tools (__api_proxy, __query_proxy, __stage_proxy, __paginate_proxy, __graphql_proxy, __sparql_proxy) lack user-facing descriptions and are not intended for direct LLM invocation. They should be hidden from the tool list or clearly marked as internal. Exposing them creates confusion about which tools agents should actually call.
No output schemas documented for any tool. LLMs cannot plan downstream calls or extract required fields without knowing what each tool returns. The search tool returns 'matching endpoints with full parameter docs' but the structure is not specified.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | F | 36 | 2026-07-28+ | v2 |
Internal proxy handler for exhaustive pagination. Implements api.getAll() to fetch all pages of a paginated API endpoint and auto-stage large result sets.
Internal proxy handler for querying staged data via SQL. Executes SQL queries against auto-staged datasets stored in Durable Objects SQLite.
Internal proxy handler for SPARQL queries executed within the isolate. Routes SPARQL requests through the host's sparqlFetch adapter with auto-staging for large result sets.
Internal proxy handler for manually staging data into Durable Objects SQLite. Accepts data and creates queryable tables.
List files matching a glob pattern in the virtual filesystem (when fsDoNamespace is configured). Available in isolate code.
Read a file from the virtual filesystem (when fsDoNamespace is configured). Available in isolate code.
Write a file to the virtual filesystem (when fsDoNamespace is configured). Available in isolate code.
Code-execution tools (iedb_execute, iedb_execute GraphQL, iedb_execute SPARQL) accept arbitrary JavaScript/SPARQL without input validation or constraints. Descriptions mention 'sandboxed V8 isolate' but provide no guidance on what code is safe, what APIs are available, or what errors to expect. No error recovery guidance.
Tool naming uses underscores for internal proxies (__api_proxy, __query_proxy) but does not follow verb_noun convention. Public tools (iedb_search, iedb_execute) lack clear action verbs, 'execute' is vague (execute what? code? a query? an API call?). Ambiguous names reduce LLM clarity.
fs.readFile, fs.writeFile, fs.glob descriptions are minimal (50 chars) and do not explain when to use them, what filesystem they access, or what errors occur. Parameter descriptions are terse ('File path', 'Glob pattern'). No guidance on path constraints or security implications.