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

Annotations only say it is a non-destructive, non-read-only, closed-world write. The description adds real behavior: creating a collection also auto-creates a tag and a draft workflow bound to that tag, and get_collection exposes them via automation.tag and automation.workflowPath. That side-effect disclosure is the kind of context annotations cannot carry. It does not describe failure modes or what happens when requestId is reused with a different config, which keeps it at 4.

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 core creation semantics are front-loaded, which is good, but the later sentences pile on cross-references (describe_collections, get_collection, automation.tag, automation.workflowPath) and the phrase 'Ouvrir ce lien' refers to a link that is never named, forcing the reader to infer it means workflowPath. A mid-paragraph switch of subject without clear transitions costs readability.

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 tool with a fully untyped nested config and no output schema, the description covers the important ground: idempotency, config source-of-truth, automatic tag/workflow creation, and the post-creation activation step. The gap is that an agent still cannot construct config without calling describe_collections, but the definition explicitly says so, which is an acceptable handoff rather than an omission.

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 0% and config is an unconstrained nested object, so the description has to compensate on its own. It does add meaning — requestId is a stable UUID enabling idempotent creation, and config is said to hold columns, sources, mode, limit and workflow rules per describe_collections — but it does not give field names, types or defaults, leaving the actual config shape entirely delegated.

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 gives a specific verb and resource ('Créer une collecte') and enumerates what the creation covers: columns, sources, auto/api/browser mode, limit and workflow rules. It is clearly distinguishable from update_collection, delete_collection and create_record. It stops short of explicitly contrasting itself with those siblings, so a 4 rather than a 5.

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?

There is useful implied sequencing: 'config selon describe_collections' tells the agent to fetch the config shape first, and the closing sentence says to open the workflowPath link and then activate the workflow. However, it never states when to use this tool versus update_collection or prepare_collection_automation, nor any prerequisites such as required permissions or a workspace scope.

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