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

No annotations are provided, so the description carries the full transparency burden. It discloses critical behaviors: the endpoint refuses 400 on construction_critique, the auto path detection (walls → flood fill → room centres), that an edited chemin is used verbatim and checked, the enrichment behavior for long durations, and that rendering is done by one browser-GPU engine. It doesn't explicitly describe the return format beyond naming the fields, but the output schema is absent so it does enumerate plan structure. Strong coverage given no annotations.

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 is dense and front-loaded with the key purpose and critical prerequisite, which is good. However, it is a sizeable paragraph mixing prerequisites, output-format documentation, consumption instructions, adjustment advice, and endpoint references — an agent must parse through a wall of text to extract the usage contract. The structure is linear prose rather than clearly sectioned, though it's organized enough to weigh somewhat more favorably than a bullet-dump.

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?

Despite lacking an output schema, the description documents the return shape ({plan:{objectif, cadence, plans[], chemin[]}, alertes[], resume}) and the chemin coordinates convention (position/vise in mm, CAO axes). For a complex tool with 17 parameters and rich geometry (objectif, cadence, chemin, toitures, reseaux), the description covers the operational prerequisites, failure modes, fix workflow, and consumption path thoroughly. It's complete for accurate invocation.

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 82% and quite rich, so baseline is high. The description adds real value beyond the schema on several key params: it names the lens catalogue endpoint (GET /api/v1/cao/visite/objectifs), explains the marche interior walk threshold (~1600 mm/s unwatchable), and details duree_s behavior (path ENRICHED with exterior revolutions rather than slowed down). The chemin field is well-explained (verbatim + checked). Slight deduction because some scalar params like resolution, fps, mode are left to the schema alone, but overall the description meaningfully supplements schema info.

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 vivid, specific framing of the tool's purpose: directing the 4K cinematic visit of a CAO scene, choosing lens and camera path. It clearly distinguishes from siblings like cao_visite_camera and cao_rendu by specifying this is a motion-planning cinematic endpoint. The verb+resource (DIRECT a 4K cinematic visit) is precise and distinct from the sibling tools.

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?

Excellent usage guidance. It explicitly states a mandatory prerequisite (call cao_verifier FIRST and fix every severite=critique), explains the 400 construction_critique failure reason, prescribes the exact fix endpoint (POST /api/v1/cao/integrite with params), and tells how to retry. It also gives adjustment instructions (resend with modified objectif/cadence or an edited chemin), plus the output consumption path (paste into CAO editor or window.CAD_startCinematicTour). This is unusually thorough.

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