Hand off an in-flight task to a human operator with a full context bundle: transcript, prior actions, identifiers, and a recommended next step. [$0.2/per_call] [async→get_outcome]
| Name | Required | Description | Default |
|---|---|---|---|
| reason | Yes | ||
| smb_id | Yes | ||
| context | Yes | ||
| priority | No | normal | |
| idempotency_key | No | Retry key: a 24h replay returns the original receipt, not re-run or charged. |
Changes observed during successful MCP inspections.
Input schema / properties / idempotency_key / descriptionPrevious value: -"Optional client-supplied key for safe retries. Replaying the same key within 24h returns the original receipt - the operation is NOT re-executed and NOT re-charged."New value: +"Retry key: a 24h replay returns the original receipt, not re-run or charged."Input schema / properties / idempotency_keyAdded value: +{
+ "description": "Optional client-supplied key for safe retries. Replaying the same key within 24h returns the original receipt - the operation is NOT re-executed and NOT re-charged.",
+ "maxLength": 128,
+ "type": "string"
+}Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already provide destructiveHint=trueasi, so the description does not need to restate irreversibility. The description adds useful behavioral context beyond annotations: the cost ($0.2/per_call), the async pattern (async→get_outcome), and the exact contents of the context bundle (transcript, prior actions, identifiers, recommended next step). This is strong added transparency.
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 core description is a single efficient sentence that front-loads purpose and then packs cost and async behavior into compact brackets. There is no fluff, and every part earns its place.
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 context, five parameters, and no output schema, the description is adequate but incomplete. It mentions async→get_outcome but does not explicitly say what the call returns or how the escalation outcome will be surfaced. The reason enum and idempotency_key description in the schema fill some gaps, but the description alone does not fully cover the call's end-to-end behavior.
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 only 20%, so the description partially compensates by naming the context-bundle fields: transcript, prior actions, identifiers, and recommended next step. It does not explain required parameters like smb_id, reason, or priority, though their names and enums are reasonably self-explanatory. Overall it adds some meaning but does not fully cover the low schema coverage.
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 uses a specific verb ('Hand off') and resource ('in-flight task to a human operator'), and explains that a context bundle is included. This distinguishes it clearly from sibling tools like send_message or get_outcome, and it is not a tautology of the name.
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 implies this is for in-flight tasks and points to get_outcome for follow-up. However, it does not explicitly state when to use this vs. alternatives like handle_inbound or send_message, nor does it state when not to escalate. The usage context is mostly implied rather than prescribed.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Add one secure layer between your agents and this server.