train_dryrun

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

No annotations exist, so the description carries the full burden. It does disclose several key behaviors: running a minimal-sample pipeline check, sampling data quality, estimating VRAM with early warn/warning on over-limit, and extrapolating cost with pre-submission budget warnings. It also references the sigmoid scaling law. However, it does not state whether the dry-run has side effects, whether it returns a report or structured result, or whether any training is actually launched. The name 'dryrun' implies safety, but explicit side-effect disclosure is missing.

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 a single well-structured sentence that front-loads the primary purpose ('训练 dry-run') and then lists the four checks in a compact, parallel format. Every clause contributes useful information; there is no fluff or redundancy. It is concise given the complexity of the tool.

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 tool with nested objects sequences, no output schema, and no annotations, the description provides a good high-level overview but lacks specifics that an agent may need: what does the dry-run return (e.g., a report? warnings?), any preconditions (e.g., data format, column mapping defaults), and whether the tool performs any mutation. The schema covers parameter details, but the description does not fully complete the behavioral picture on its own.

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 100%, so the baseline is 3. The tool description provides high-level context by mapping its checks to parameters (e.g., '显存估算' maps to the vram object, '算力外推' maps to extrapolate), but it does not add new per-parameter semantics beyond what the schema already states. It is sufficient but not additive.

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 clearly identifies this as a pre-submission dry-run for training, distinct from actual training submission. It enumerates four specific checks (pipeline connectivity, data quality sampling, VRAM estimation, compute extrapolation) that precisely define the tool's scope. The name train_dryrun aligns with the description and differentiates it from siblings like train_submit.

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?

The phrase '提交前预检' (pre-submission precheck) establishes clear context for when to use this tool: before a training job is submitted. It implies a pre-submission workflow but does not explicitly name alternatives or state when not to use it. No exclusions are given, but the usage context is unambiguous.

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