dsh_send

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

No annotations exist, so the description carries the full burden and does so: it discloses that a timeout does not error and does not cancel the turn but degrades to background, that results land atomically in a sentinel file, that reasoning is hidden by default with a context-cost warning for `full`, and that a busy conversation produces an error. These are non-obvious failure-mode behaviors an agent could not infer from the schema.

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 default non-blocking behavior is correctly front-loaded, but the same point — that timeouts never destroy the task — is made three separate times with heavy bolding, and the result-retrieval mechanics appear twice. Substantial length is spent on emphasis rather than new information.

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 dispatch tool with no annotations and no output schema, the description covers everything needed: what is returned (run_id + sentinel_file), how to obtain the final result (sentinel file, dsh_get, dsh_read mid-run), default reasoning visibility, and the error path. An agent has no unanswered question before calling it.

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% and the per-parameter descriptions already cover the wait default, timeout_ms degradation semantics, reasoning levels, and conversation_id provenance. The prose largely restates those same facts rather than adding new meaning, so the baseline 3 applies; the only marginal addition is the cross-parameter framing that timeout_ms matters solely when wait=true.

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 opening sentence names a specific verb and resource (dispatch a task to a conversation) and immediately distinguishes the tool's two modes. It also names the siblings it is not for the busy case (dsh_interject, dsh_interrupt) and the verification path (dsh_get, dsh_read), so an agent can place it in the family without opening other schemas.

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?

Explicit when-to-use: default wait=false for fire-and-forget, explicit wait=true to block in place, and a recommended Bash run_in_background + `until [ -f sentinel_file ]` pattern instead of polling. It also states the negative case — if the conversation is busy the tool errors and you should use dsh_interject or dsh_interrupt — which is exactly the alternative routing an agent needs.

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