Fetch a sealed candidate record (a narrowing receipt or a commitment anchor) by its content hash — the whole preserved path, losers included, re-checkable offline with tools/verify_seal.py. The seal proves process integrity, never truth.
| Name | Required | Description | Default |
|---|---|---|---|
| hash | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
No annotations are provided, so the description carries the full burden. It clearly discloses that the response contains the 'whole preserved path, losers included', that verification is possible offline, and that the seal does not establish truth. This is rich behavioral context, though it stops short of mentioning error behavior or side effects (though 'fetch' implies read-only).
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is concise, with the primary action and object in the first sentence, and a clarifying second sentence about semantics and verification. No filler or redundant information.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a simple one-parameter fetch tool with no output schema, the description covers the return content, the verification workflow, and the interpretive caveat. This is enough for an agent to select and invoke the tool correctly.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema description coverage is 0%, so the description must explain the 'hash' parameter. It does so by calling it a 'content hash', which tells the agent it is a hash of the record content. This is sufficient for a single simple parameter, but adds little beyond what the schema pattern already implies.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description opens with a specific verb+resource: 'Fetch a sealed candidate record' by content hash. It further clarifies the record types ('narrowing receipt or a commitment anchor') and what the response includes ('the whole preserved path, losers included'), distinguishing it from siblings like candidate_commit or candidate_narrow.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies usage context by stating the record is 're-checkable offline with tools/verify_seal.py' and warns that the seal 'proves process integrity, never truth'. However, it does not explicitly state when to use this tool over alternatives like seal_fetch or verify, nor does it provide exclusionary guidance.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Add one secure layer between your agents and this server.
Several tools are near-duplicates: read_passage and resolve both fetch WEB text for a reference; word_study already includes every occurrence that word_occurrences returns; coach_next and coach_recommend both answer 'what's next.' Search/locate/cards_browse also overlap as discovery entry points, making tool selection ambiguous despite detailed descriptions.
Most names follow an object_verb snake_case pattern (cards_browse, study_create, seal_fetch), but there are many bare verbs/nouns (ask, audit, resolve, verify, canon, harmony) and inconsistent singular/plural pairs (card_get vs cards_browse, group_create vs groups_list, want_open vs wants_list). No camelCase, but the convention is not uniform.
86 tools is an extreme count for any single MCP server, far beyond the 3-15 well-scoped range; even a broad platform would be hard for an agent to navigate. Many tools belong to unrelated subdomains (coach, steward, mesh, calendar), making the surface unwieldy.
The want/offer flow has no accept/close tool, so an agent can open a want and offer a source but never see it resolved. Group and calendar coverage are one-directional (create/join only; no leave/delete/list/update), and there is no badge listing or way to update a study group. Core reading/verification/shelf flows are solid, but lifecycle gaps remain.