MCP server for local email archives — search and read Google Takeout .mbox exports and .eml files entirely on your machine. No OAuth, no IMAP, no cloud.
Three well-defined tools with clear, actionable descriptions (150 - 250 chars each). All tools have complete input schemas with typed parameters and descriptions. Tool names follow verb_noun convention (open_archive, search_messages, get_message). Parameters are well-constrained (date regex, numeric limits, enums implicit via descriptions). Output schemas are documented via code comments and JSON structure. Main gaps: no tool annotations (readOnlyHint/destructiveHint), no error recovery guidance in descriptions, and parameter descriptions could be more explicit about constraints (e.g., 'YYYY-MM-DD format' stated but not in schema pattern for all date fields).
Fully parse one message by its id (from search_messages): headers, decoded text body (truncated if huge), and attachment names/types/sizes. Attachment contents are never returned.
Index an .mbox file (Google Takeout, Thunderbird) or a directory of .eml files — entirely locally, nothing leaves the machine. Returns message count, date range, and top senders. Large archives are indexed once and cached.
Search indexed envelopes (from/to/subject/date). `query` matches sender and subject. `body_query` additionally full-parses candidate messages to search body text — it is capped, so narrow with sender/date filters first; the result reports how many candidates were scanned.
No tool annotations (readOnlyHint, destructiveHint, idempotentHint) declared in tool registration. All three tools are read-only and idempotent, but this is not signaled to the client.
Error descriptions lack recovery guidance. When a message ID is invalid or archive path is missing, errors state the problem but do not suggest next steps (e.g., 'Call search_messages first to get valid IDs').
Parameter 'body_query' description does not explicitly state the BODY_SCAN_CAP=500 limit or advise narrowing with filters first. Users may not understand why body search is slow or incomplete.
Output schema for search_messages includes 'bodiesScanned' and 'scanCapped' only conditionally. Documentation should clarify when these fields are present vs. absent.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 61 | 2026-07-28+ | v2 |
Date parameters (after, before) use regex validation in schema but descriptions do not explicitly state 'YYYY-MM-DD format required', relying on regex alone is not LLM-friendly.