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

Annotations already mark the tool readOnly and non-destructive, and the description adds substantial behavioral context beyond that: lean result shape, compact text-line rendering vs full object, suppression of conflict pairs, memory_id being silently ignored in other modes, and which filters apply to which modes. This gives the agent a realistic model of how the tool behaves at call time.

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 long but information-dense, with the core purpose front-loaded and each mode and filter earning its place. Some parenthetical asides, such as the shadow-rows explanation, add necessary context. It is structured effectively, though it could be tightened without losing key details.

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?

Given seven optional parameters, no output schema, and multiple modes, the description covers most of what an agent needs: mode semantics, filter behavior, output format variations, suppression behavior, and cross-tool safety guidance. Minor gaps remain, such as domain/limit semantics and definitive mode enum values, but overall it is nearly complete for a tool of this complexity.

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 only 14%, so the description carries most of the burden, and it largely succeeds: tags, node_kind, state, and memory_id all receive semantic explanations. However, domain and limit are named but not given semantics such as allowed domain values or default/maximum limits, so compensation is strong but not complete.

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 opens with a specific verb and resource: 'Inspect the health of the knowledge graph,' then enumerates four distinct modes (stale, orphans, archived, conflicts) that make the tool's scope concrete. This clearly separates audit from sibling retrieval tools like recall and search, even without naming them.

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?

Each mode is tied to a specific use case, e.g., stale surfaces drift candidates, orphans returns memories with no connections, and conflicts surfaces semantically close pairs. It also gives an explicit safety rule for archived mode ('only call forget(restore=true) ... after explicit unambiguous user confirmation') and points to recall(id) for full content. However, it does not explicitly say when to prefer a sibling tool over audit, so it stops short of full when-not guidance.

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