MCP server for Google Finance's internal batchexecute RPCs. Fetches Google Finance beta page RPC mappings from AF_initDataKeys and AF_dataServiceRequests, honors the page cache headers, and calls the Google Finance batchexecute RPC endpoint.
Google Finance MCP has well-structured tool definitions with consistent naming (verb_noun pattern: list_, call_, get_, refresh_) and complete JSON schemas for all 11 tools. Descriptions are present and contextual (avg ~120 chars), explaining what each tool does and when to use it. However, parameter descriptions lack depth, many parameters like 'request_override' and 'request' are documented only as 'Any JSON value accepted by Google Finance' without guidance on structure, constraints, or examples. Output schemas are not documented; LLMs cannot predict response structure. Error handling is absent, no guidance on retryability, user-fixable errors, or recovery paths. The 'request_override' and 'request' parameters accept arbitrary JSON with no validation hints, inviting hallucinated payloads. Tool composition is sound (each tool has one responsibility), but parameter relationships (e.g., how dataset_key maps to RPC structure) are undocumented.
Call multiple Google Finance RPCs in one batchexecute request.
Call any Google Finance home-page dataset key advertised by AF_dataServiceRequests.
Call any dataset key advertised by a specific Google Finance page path or URL.
Call a quote-page dataset key for a ticker and exchange.
Call a Google Finance RPC ID with an explicit JSON request payload.
Call the Google Finance quote Holdings tab RPC for a ticker and exchange. Returns raw data plus best-effort labels for politician holdings, politician transactions, insider transactions, and pagination cursors. This RPC is not advertised in AF_dataServiceRequests.
Undocumented output schemas. No tool documents what fields or structure the response contains. LLMs cannot predict downstream field availability or plan chained calls. Agents must guess at response structure.
'request_override' and 'request' parameters accept arbitrary JSON with only generic description 'Any JSON value accepted by Google Finance'. No guidance on structure, required fields, or valid payloads. Invites hallucinated requests.
No error handling guidance. Tools do not document what errors can occur, whether they are retryable, or what the LLM should do next. A failed RPC call returns no recovery path.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 58 | 2026-07-28+ | v2 |
List the reference RPC ID to purpose mapping from the reverse-engineering article. These IDs are examples only; live calls use the current id from AF_dataServiceRequests.
List AF_initDataKeys and AF_dataServiceRequests for a Google Finance page path or URL.
List quote-page AF_initDataKeys and AF_dataServiceRequests for a ticker and exchange.
List the current Google Finance beta AF_initDataKeys and AF_dataServiceRequests mapping, enriched with RPC purpose metadata where known.
Refresh or revalidate the Google Finance AF_dataServiceRequests mapping.
Parameter relationships undocumented. 'dataset_key' parameter appears in multiple tools but its valid values depend on the page/quote context. No guidance on how to discover valid keys or what happens if an invalid key is passed.
No pagination or result-limiting guidance. Tools like 'google_finance_list_rpcs' and 'google_finance_list_known_rpc_purposes' do not document result size, whether pagination is supported, or how to handle large result sets.