Forget a Review

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

The description asserts 'Idempotent.' while annotations declare idempotentHint=false — a direct, machine-visible contradiction on the one trait that determines whether a retry is safe. It also leaves out what a delete actually does to dependent state (the schema hints that recall compares revisions to mark records stale, but the description never says forgetting invalidates anchors). With destructiveHint=true and openWorldHint=true already in structured data, the description needed to add caution, not contradict it.

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: action, the routing rule, then the safety gate — a sensible front-loaded order with no filler. The only wasted sentence is the false idempotency claim, which costs clarity rather than length.

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?

An output schema exists, so return values needn't be described, and annotations carry the destructive/open-world profile. But the description never mentions the required `artifact` (revision + format) that every invocation must supply, and its misleading idempotency claim leaves an agent unable to decide whether a retry is safe — a material gap for a destructive mutation.

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?

This is where the description earns its keep: it explains the routing logic between `name` (named feature), `entity_id` (anonymous notes) and `subject` (decision), which the flat schema states only in passing and never as a choice hierarchy. Schema coverage is already 75%, so the baseline is 3; the explicit either/or mapping lifts it. It omits that `artifact` is required regardless of which of the other three you pass.

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 opening verb-phrase 'Delete a saved record' is then made specific by enumerating the three removal targets (named feature, entity's anonymous notes, decision), each mapped to a different argument. Siblings (remember/recall/remember_decision) are clearly a different operation class, so the tool is easily distinguished. It stops short of 5 because the mandatory `artifact` anchor — needed on every call — is never mentioned.

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?

'Use only when the user explicitly asks to remove it' gives a real gating condition rather than generic context, which for a destructive tool is the most important guidance. No alternative is named (e.g. 'use remember to overwrite instead of forgetting first'), so an agent must infer whether an update path exists.

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