recall

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

No annotations are present, so the description carries the full disclosure burden, and it does real work: it reveals search order (project dataset first, then user dataset), the presence of evidence/validity/superseded/contested flags on facts, and the explicit 'you synthesise' instruction that tells the agent not to expect a final answer. It also discloses automatic mode routing unless overridden. It stops short of documenting pagination, cost/latency, or how hops/limit shape results, but the disclosed traits are substantive and non-obvious.

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?

Three sentences with no filler; the capability, mode enumeration, search scope, and return contract are all packed in economically and front-loaded. Every sentence contributes distinct information, including the short 'you synthesise' tip that prevents a common misuse. The mode list is dense but earns its place because the schema defines no enums for `mode`.

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 7-parameter tool with no annotations and no output schema, the description covers routing, search scope, and return format well, which is the core of correct invocation. Missing semantics for `hops` and `limit`, and the lack of guidance on how `datasets` interacts with `include_user`, are real gaps an agent would hit when tuning retrieval. It is adequate for a first call but incomplete for sophisticated filtering.

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 compensate; it fully documents `mode` by enumerating all seven values plus the default, and it illuminates dataset-scoping parameters (`include_user`, `datasets`) through the 'searches the project dataset then the user dataset' statement. The `query`, `hops`, and `limit` parameters receive no meaning beyond their schema names, which is a notable gap given zero schema descriptions. Overall it adds substantial value for the most complex parameter while leaving others unexplained.

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 names the verb ('Recall') and resource ('memory'), then specifies the routing modes (hybrid, facts, neighbourhood, lexical, summaries, temporal, rules, session), the search order (project then user dataset), and the return types (ranked raw entities, facts with evidence/validity/superseded/contested flags, chunks). The mode list and return-type disclosure clearly distinguish it from retrieval-adjacent siblings such as `history`, `global_context`, and `session_get`. The opening phrase is mildly tautological, but the following detail makes the purpose unmistakable.

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 mode list implicitly signals when each behavior applies — e.g., 'lexical (quoted phrase)' tells the agent that quoted phrases select lexical matching — and the 'unless given' phrasing clarifies that mode is otherwise auto-routed with a stated default. However, there is no explicit when-to-use vs when-not-to-use guidance, no mention of alternatives like `history`, `session_get`, or `global_context`, and no exclusions. Usage context is implied rather than deliberately stated.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.