training_backends

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

With no annotations, the description carries the full burden, and it does disclose the critical non-mutation guarantee ('nothing is downloaded') plus the fact that it reports host-level incompatibilities like ROCm on Windows/RDNA2. It stops short of covering permissions, cost, or execution time, but the safety profile is clear.

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?

Two dense sentences that front-load the core question ('which training stacks exist here') before the caveats. The stack enumeration is long but each item is material to what the tool reports, so little is wasted; only the comma-spliced list slows reading slightly.

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?

With no output schema, the description compensates by describing the return content: install status, install lines, and unsupported-host notes. That is sufficient for a zero-parameter diagnostic call, though it does not specify the result format (structured vs. prose) an agent will have to parse.

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?

The tool takes zero parameters, so there is nothing for the description to disambiguate; the baseline applies. The enumerated stack list does help the agent know what scope of answer to expect, even though it is not a parameter.

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?

Names the concrete resource (installed training stacks) and enumerates exactly which ones it covers, including torch/ROCm, AITER, JAX, TorchTitan and others. The closing clause 'Detection only; nothing is downloaded' pins down the operation as a read/query rather than an install, so an agent can distinguish it from install_opencode_plugin or export_training_data at a glance.

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 'Detection only; nothing is downloaded' implicitly tells the agent when this tool is appropriate (checking availability before training/installing), but there is no explicit when-not guidance or named alternative such as doctor for general host health. Usage is inferable rather than stated.

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