shell_bulk

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

No annotations are provided, so the description carries the full burden. It does disclose meaningful behavior: in-order sequential execution, per-step retry/condition/failure handling, cross-step state reading (prev.*, steps[i], vars.<name>, failed_count), and token-cost tradeoffs via capture. It omits the risky part of the picture: no warning about destructive side effects, no statement of permissions or sandboxing, and no hint about the shape of what comes back. Adequate but incomplete for a shell-execution tool with zero annotation coverage.

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 sentences, front-loaded with the core purpose and then the step capabilities and expression namespaces. Dense but readable, with no filler prose. The middle list partially mirrors schema fields, which is slight redundancy, but it earns its place by framing the features before the agent opens the nested schema.

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 10-parameter tool with nested step objects, no output schema and no annotations, the description covers the essential mental model: batching, ordering, per-step conditions, retries, failure policy, variable assignment, and output capture. What is missing is the return structure of a run and explicit positioning against the other shell siblings, but the condition namespaces (prev.ok, failed_count) hint at the result shape.

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 description still adds value beyond the schema: it enumerates the readable namespaces for conditions and ${...} interpolation (prev.ok, prev.exit, prev.stdout, step.<id>.ok, steps[0].exit, vars.<name>, failed_count), which the schema only gestures at with an example. It also paraphrases assign and capture succinctly, so an agent can grasp the semantics without reading every nested property.

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 and resource with a distinguishing scope: 'Run MANY commands in one call, in order', which implicitly separates it from the single-command siblings like shell_exec. It also tags the value proposition ('token-efficient way to work'). It stops short of naming an alternative tool outright, so the differentiation is inferred rather than stated.

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 description gives one concrete usage preference ('Prefer capture="on_failure" for long pipelines') with the reason why (silent on success, full output where it broke). However, it never says when to choose shell_bulk over shell_exec or shell_exec_async, nor when sequential batching is inappropriate. Usage is implied by the batching framing rather than asserted.

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