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

Annotations already declare readOnlyHint=true and destructiveHint=false, so safety is covered; the description still adds real value by stating "Nothing runs and nothing is charged" (i.e. compilation is free and non-executing) and by explaining that ast_hash identifies this particular reading. The retained-mode behavior — the server reloads the prior Spec and callers cannot substitute a base — adds semantics no annotation conveys.

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 the core action and the return shape before moving to the retained variant, and no sentence is filler. Some sentences are long and densely comma-listed (the domain enumeration), which costs a little scannability but not correctness.

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?

With no output schema, the description correctly spends words on the return payload (find target, conditions, required/optional facts, count, order, ast_hash), and it covers the single nested object's purpose. The main gap is the unmentioned locale parameter on a tool explicitly about multi-language input, which is the one place an agent could make a wrong call.

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 reported as 0%, so the description has to carry the load. It handles q (free-form request in any language) and retained well, explaining that add-facts/extend/refresh keep the original target and conditions and only re-read new facts or the new count. It says nothing about the locale parameter or the ast_hash 64-hex format, leaving one of three top-level parameters 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?

"Reads a request for a list, written in everyday words in any language" plus an enumeration of the domains it spans (companies, jobs, papers, trials, datasets, ...) and the shape of what it returns gives a concrete verb+resource. The opaque tool name 'compile' is fully redeemed by the text, but it never names a sibling (create_spec, edit_spec, spec_options) that an agent might confuse it with.

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?

It states the condition for the retained path clearly: when the user already has a list, supply parent_find_id and an operation. It also states the precondition ("The caller needs read access to that list"). What is missing is any exclusion — nothing says when to prefer create_spec, edit_spec or spec_options instead, so routing between siblings still requires inference.

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