khwan_recall

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

Even without annotations, the description discloses key behaviors: the server ranks and enforces a top-3 ceiling, a relevance floor can cause empty facts which is a valid answer, lessons lead the seed text, and no model is called. It explicitly mentions that limit can only narrow not widen, and that seed_text can be empty. This is far beyond what annotations would normally provide.

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 longer than typical but each section earns its place: the opening sentence states purpose, the constraints are genuinely important and specific to the tool, and the returns section aligns with the schema. However, the call-to-action highlighting bold and paragraph spacing adds length; the seed_text return description is slightly redundant with 'facts' and 'lessons' in the returns listing. There is a small amount of wasted space around the returns details, but overall it is structured and scannable.

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?

Despite having no formal annotations, the description thoroughly covers when to call, what it returns, what the limits are, how to interpret empty facts, and how to phrase the query. The output schema fills in the structure details, and the description adds semantics about the returned lessons leading the seed, the count, and seed_text. For a memory tool with no annotations, this is more than complete enough for an agent to invoke it safely and effectively.

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 coverage is 0% in this context, so the description carries the full burden. It actually does that: query is defined as a meaningful task description matched on meaning, and limit is described as a cap between 1-3 and can only narrow. That is strong semantic enrichment of the input schema. I'd give 5, but the description doesn't specify the default or required behavior of limit beyond 'leave it alone unless...'—though it implicitly references the default 3. Minor gap relative to perfect.

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 says explicitly that this tool SEEDs a session/subagent with a compact bounded set of relevant memories, names the exact resource (relevant memories/facts) and the verb (seed/recall). It distinguishes itself from siblings by positioning itself as the token-smart entry point for recall at session start, not per-turn, and by noting that no model is called. It is neither tautological nor vague.

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 gives direct instructions: call ONCE at session start or when a fact scrolled out of context, NOT on every turn. It also explains when to avoid retrying on empty results and warns about the relevance floor. It doesn't explicitly name an alternative sibling, but it clearly defines the expected invocation context and anti-patterns, which is stronger than merely naming a sibling.

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