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

Annotations only give the generic mutation profile (readOnlyHint=false, destructiveHint=false, idempotentHint=false). The description goes well beyond: it discloses the additive/idempotent-ish semantics (exact repeat reinforced, anything else creates new), that close matches surface in `similar` and are never overwritten, that `supersedes` only marks rather than deletes, and two concrete refusal conditions. This is exactly the extra context annotations cannot carry.

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?

Front-loaded with purpose and the trigger condition, then a compact scannable parameter list, then the additive/refusal notes. Dense but nearly every sentence carries signal; the trailing refusal sentence is slightly list-like but justified given the security implications.

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 9-parameter, no-output-schema write tool, the description covers the important behavioral surface and even names relevant return keys (`similar`, `warnings`). The only real gap is the unexplained `force` parameter and any detail on the shape of confirmation returned, which is minor.

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 carry parameter meaning, and it does for 8 of 9 params (content, type with full enum, project, tags, importance, on_similar, status, supersedes) including non-obvious semantics like status=proposal vs decision. The one omission is `force`, which is undocumented anywhere, keeping this just short of full coverage.

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?

States a specific verb and resource ("Saves a memory") and enumerates the content kinds it handles (note, decision, preference, to-do), so an agent immediately knows this is the write path. It never names a sibling to contrast against, but no sibling in the list also writes, so disambiguation is effectively implicit.

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?

Explicitly prescribes when to call it: "Call it on your own, without asking first, whenever the user mentions something about themselves or their work worth keeping, even in passing." It also states when NOT to expect success (private mode `##` prefix, secrets/card numbers/passwords) and how to handle near-duplicates via `on_similar`. That is real when/when-not guidance, not inference.

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