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

Annotations only declare readOnly/destructive/idempotent/openWorld flags; the description goes well beyond them by stating 'The caller remains responsible for enforcement,' that missing evidence must be collected rather than fabricated, that the result is not an authorization, and that there is no truth/legal/compliance verification and human review is required. Those are exactly the non-obvious traits an agent needs. It is slightly undermined by the large promotional block that muddles the operation's actual behavior.

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 routing purpose is front-loaded, but roughly half the text is unrelated sales material: five free trial attempts, MCP free_trial/free_trial_status calls, an HTTP trial URL, Base USDC payment requirements, X-Payment-Signature retention, and a pricing link. That is noise for an agent deciding whether to invoke this tool and buries the operational guidance.

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 need not be described, and the description covers the input contract, the resubmission semantics, and the non-authorization disclaimer. For a tool with 7 required parameters and nested claim/evidence structures this is reasonably complete, though it does not explain what each routing outcome means in practice, which would help an agent interpret the result.

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 10% across 10 parameters, so the description has to carry the load, and it enumerates all seven required arguments plus the optional approval reference and policy checks. The embedded synthetic example is the strongest contributor, showing the shape of actor, target, claims/support_level, and evidence objects that the schema leaves undocumented. It still says nothing about unsupported_claims or the internal structure of policy_context.checks, so it does not fully close the coverage gap.

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 sentence states a specific verb and resource with concrete outcomes: 'Route a caller-controlled action to continue, request_evidence, require_approval, or stop using supplied claim evidence, risk tier, policy checks, and an optional external approval reference.' That is far more than a restatement of the name. It loses a point because it never differentiates itself from the sibling tools decision_check and decision_verify, which sound like they operate in the same decisioning space.

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 gives a clear trigger ('Before an action, supply the actor, intended action, target, risk tier, actual claim evidence...') and a workflow rule for follow-up ('Resubmit the same run_id after adding evidence or approval'). It also sets a boundary on behavior ('do not invent it or treat the result as authorization'). What is missing is any when-not guidance or routing to alternatives such as decision_check or free_trial.

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