submit_optimization

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

Adds substantial context beyond the annotations: the job runs in an isolated copy, returns promptly with a run_id, sends selected code to StepFun, may incur API charges, and never mutates the parent model or source. The egress/cost disclosure is exactly the kind of behavior an agent needs before invoking, and it is consistent with destructiveHint=false.

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?

Four tight sentences, no repetition, and front-loaded: what it starts, what it requires, what it returns, and the cost/safety implications.

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 fire-and-forget submission tool with no output schema, the description adequately covers lifecycle (run_id + poll), cost, and non-destructiveness. The gap is that a very complex nested parameter object is left largely unexplained, which an agent configuring a benchmark job would need.

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% on a deeply nested JobSpec, so the description must carry the load. It names several required inputs (source_dir, files, checks, benchmark, budget), but leaves editable_files, direction, metric, repetitions, constraints, final_checks, and the entire Budget knob set undocumented in both places.

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?

States a specific verb (Start) and resource (a Step API optimization job in an isolated copy), and implicitly distinguishes itself from the sibling status/result/cancel tools by describing a one-shot submission that returns a run_id to poll.

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?

Prerequisites are enumerated clearly ('Requires an allowed source_dir, explicit files, correctness checks, benchmark, and budget'), and it routes the agent to polling after submission. It doesn't name get_optimization_status/get_optimization_result explicitly, so it stops short of full when-to-use guidance.

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