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

With no annotations provided, the description carries the full behavioral burden and discharges it thoroughly: it discloses the ~2-5 min runtime with no mid-call progress, the degraded:true failure signature (degraded_reason=assessment_parse_failure with an empty findings list) triggered by oversized diffs, the 1 MB input clipping notice, and the return shape (structured critique with severity tags plus a recommended action class). This is effectively a self-contained annotation layer.

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 runs long but every sentence is operational, and it is tightly sequenced: triggering scope, mechanism, return value, timing with a user-facing instruction, input fidelity requirement, size-limit handling, degradation signature, then alternative routing. Information is front-loaded (the when-to-use scoping comes first) and the length is justified by the tool's complexity — 12 parameters, a long runtime, and multiple failure modes.

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 high-complexity tool with no output schema and 0% schema coverage, the description covers the core workflow well — what it returns (critique with severity tags, action class), when it degrades, how long it runs. The critical gap is parameter breadth: 11 of 12 parameters remain semantically unexplained, and no output schema exists to compensate. The main agent-facing path (passing a diff) is complete, but the broader parameter surface is not.

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% across 12 parameters, so the description must compensate — but it only documents one: `diff`. That parameter is handled excellently (pass the REAL text, never a summary; split past 50 KB; clipping behavior), yet the other 11 parameters (tests, gate_diff, gate_repo, constraints, eval_case_id, relevant_code, gate_context_id, proposed_action, continuation_token, target_hunk_hashes, architectural_context) have zero semantic explanation in either the schema or the description. Several are technical (continuation_token, target_hunk_hashes) and entirely opaque to an agent.

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 precise triggering condition — 'before committing to a coding decision you've drafted' — and enumerates concrete scenarios (pre-merge diff, architectural choice, security assessment, migration plan). It names the mechanism (four frontier models stress-testing for blind spots), and differentiates itself from siblings by naming the fast/open-ended alternatives it is not. An agent can select this tool over synthesize_coding and deliberate_coding without inspecting schemas.

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?

Usage guidance is explicit and operational: it states exactly when to call (before committing a drafted decision), how to behave pre-call (tell the user it's working, since no progress shows mid-call), and how to handle large inputs (split diffs past 50 KB into 2-3 chunked calls). It ends by naming the two alternatives with their selecting conditions — 'fast broad take' → synthesize_coding, 'open-ended decision' → deliberate_coding. No exclusion condition is left to inference.

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