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

Annotations already declare destructiveHint=true, idempotentHint=true and readOnlyHint=false, so the danger profile is covered structurally. The description's genuine additions are the dry-run preview behavior and the human-confirmation gate, plus the user-visible effect on agent answers. It does not say whether removal is recoverable, whether a source still attached to agents is refused or cascaded, or what the preview actually returns.

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?

Two sentences with no filler: the purpose and its consequence come first, the call protocol second. Every clause carries operational meaning for the caller.

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 destructive, zero-output-schema tool the description covers the essentials: what is affected, that a preview exists, and that explicit confirmation is required. The remaining gap is outcome reporting — with no output schema, the caller cannot tell from the description what the preview or success response contains, or how partial failures surface.

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 only 33%: source_id carries a description ('From get_knowledge_base') while knowledge_base_id and approved do not. The description compensates for the least obvious parameter by explaining that omitting 'approved' yields a preview, which the schema cannot convey. It still leaves the identifier formats and validation behavior undocumented.

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 ('Take one source out of a knowledge base') and adds the downstream consequence ('so agents no longer find answers from it'), which makes the effect of removal concrete. It is clearly separable from add_knowledge_source by the verb, though it does not explicitly contrast itself with adjacent operations such as archiving.

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?

Gives an explicit two-step invocation protocol: call without 'approved' to preview, then call with 'approved: true' after a person confirms. That is real when-to-use and how-to-sequence guidance rather than implied usage. It stops short of naming any alternative tool or stating when removal is the wrong choice.

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