Save to memory

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

The annotations only declare the safety envelope (readOnlyHint=false, destructiveHint=false, openWorldHint=false); the description adds real behavior on top: 'A near-duplicate updates the older memory' discloses upsert/dedup semantics that the agent could not otherwise know, and 'in any of their AI apps' reveals the memory is cross-application in scope. 'One self-contained fact per call' also implies an atomicity constraint that affects how an agent should batch facts. It does not say whether the write is confirmed back to the user or what happens at the 4000-character ceiling.

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, each doing distinct work: trigger, content shape, duplicate handling. The usage trigger is correctly front-loaded so an agent scanning the first clause already knows whether to call the tool. Minor looseness in 'in any of their AI apps,' which reads as slightly vague marketing phrasing rather than a precise scope statement.

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?

There is no output schema, so the agent's real gaps are when to write, what shape the content takes, and how duplicates behave, and all three are covered. With only three parameters and annotations already carrying the non-destructive safety profile, the definition is nearly self-sufficient. The unexplained `kind` enum and the silent length limit are the remaining holes.

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 67%, with the `project` parameter already documented in the schema and `content` partially documented there. The description does add format guidance the schema lacks, namely that content should be a summary first line followed by detail, and readable in isolation. However, the `kind` enum (user/feedback/project/reference/note) receives no explanation at all, which is the parameter most likely to be mis-selected.

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 states a concrete outcome (persisting a durable fact about the user across future conversations) and enumerates the fact categories: identity, preference, decision, person, pointer. The verb 'remember' plus the title 'Save to memory' makes the action unambiguous, and the opposite siblings (recall, forget) are easy to separate since this is the only write-side tool. It stops short of naming those siblings, but the purpose is never in doubt.

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 opening clause 'Use this when the user tells you something worth knowing in future conversations' gives an explicit trigger condition, and the category list ('who they are, a preference, a decision, a person, a pointer') clarifies what qualifies as worth saving. What's missing is a when-not-to-use clause or a pointer to recall/forget for the inverse operations, so the agent still has to infer that transient or already-known facts shouldn't be stored.

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