Fetch a shelved episode by ID, or just one section, to answer a question from indexed agent memory and reduce context tokens.
Fetch a shelved episode by id, or one ## Section of it, as a data
envelope. Prefer the section when it answers the question.
| Name | Required | Description | Default |
|---|---|---|---|
| params | Yes |
| Name | Required | Description | Default |
|---|---|---|---|
| result | Yes |
Changes observed during successful MCP inspections.
v0.3.0Input schema / $defs / RecallInput / additionalPropertiesAdded value: +falseInput schema / $defs / RecallInput / properties / log / defaultPrevious value: -falseNew value: +trueInput schema / $defs / RecallInput / properties / log / descriptionPrevious value: -"Append this recall to recall-log.tsv (feeds realized-economy stats)."New value: +"Append this recall to recall-log.tsv (feeds realized-economy stats). On by default (#112): without the counter there is no way to know whether the reading habit ever took; pass false to read without a trace."Input schema / $defs / RecallInput / properties / shelf_path / defaultAdded value: +""Input schema / $defs / RecallInput / properties / shelf_path / descriptionPrevious value: -"Path to an initialized memory shelf."New value: +"Path to an initialized memory shelf. Optional: when omitted, the shelf named by $MEMSHELF_SHELF_PATH is used. Pass it explicitly to address a different shelf than that default."Input schema / $defs / RecallInput / requiredPrevious value: -[
- "shelf_path",
- "episode_id"
-]New value: +[
+ "episode_id"
+]v0.1.0Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint, idempotentHint, and destructiveHint=false, so the safety profile is covered by structured data. The description adds only the vague 'as a data envelope' return-shape hint and says nothing about the default side effect of appending a recall-log entry (documented only in the `log` schema property) or about shelf resolution via $MEMSHELF_SHELF_PATH.
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?
Two compact sentences with the core fetch semantics front-loaded and no filler. The second sentence earns its place by routing the `section` choice, though it could have been folded into the first for a tighter single-sentence definition.
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?
With an output schema present, the description needn't explain the envelope contents, and read-only safety is covered by annotations. What is missing is minimal but real: no note that recalls are logged by default, which is the one behavior an agent might want to know before calling it.
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?
The visible schema itself documents episode_id, section, log, and shelf_path in some depth, so the description is not obligated to carry the full load; it reinforces the id and `## Section` selectors but says nothing about `log`, `max_bytes`, or `shelf_path`. The raw coverage signal (0%) reflects the top-level `params` wrapper, but the nested schema descriptions mean the description's marginal contribution is small.
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?
Specific verb ('Fetch') plus resource ('a shelved episode by id, or one `## Section` of it') tells the agent exactly what comes back, and 'by id' implicitly separates this from the keyword-oriented memshelf_search sibling. It stops short of naming any sibling explicitly, so the differentiation is inferred rather than stated.
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?
'Prefer the section when it answers the question' is a genuine when-to-use hint for the `section` parameter, which is the main branching decision in this tool. However, there is no guidance on when to recall by id versus using memshelf_search, memshelf_advise, or memshelf_resolve, nor any stated precondition that the episode must already have been shelved.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.