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

Annotations only declare readOnlyHint=false, destructiveHint=false and openWorldHint=true; the description adds real behavior: retry resumes at the errored step, completed steps are not replayed, and external calls may already have succeeded before a timeout. That side-effect warning is valuable precisely because destructiveHint=false could otherwise suggest a harmless replay. It does not describe the response, the failure mode when the run is not in a failed state, or any permission requirement.

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?

Three short sentences, front-loaded with the action and scope, followed by the idempotency warning and the step-replay rule. No filler and nothing repeated from the schema or title.

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 two-parameter control tool with no output schema, the essential information an agent needs before acting — including the risk of duplicating an already-successful external call — is present. Remaining gaps (response shape, behavior on a non-failed run, whether cancel is reversible) are minor but not zero, and the description is written in French while the surrounding tool surface is English.

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%, so the description must carry the parameter burden. It does explain what the two enum values mean in operation terms (cancel vs. resume at the errored step) and implies id identifies the run, but never says where id comes from (list_runs/get_run) or what form it takes.

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 names two specific verbs (cancel/annuler, resume/reprendre) and a precise resource scope: a failed run, resumed at its errored step. That is enough to separate it from read-only siblings like get_run and list_runs. It does not explicitly distinguish itself from retry_workflow_run, which is a different resource family, so sibling differentiation is implicit rather than stated.

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 context is clear: this tool is for a run that has failed, and it explains the pre-condition of checking whether an external request actually succeeded before retrying. It stops short of naming the tool to use for that verification (e.g. get_run) and does not say when to prefer cancel over retry.

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