ask_humanPause the task, ask a specific human a structured question, and put the task in review. Prefer this over guessing when a decision requires human input.
| Name | Required | Description | Default |
|---|---|---|---|
| task_id | Yes | Task id, or a pasted Tango task URL. | |
| question | Yes | The exact question you need answered. | |
| to_user_id | No | Optional recipient user id. Defaults to the task's approver or owner. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Adds important side effects beyond the annotations: the task is paused, a specific human is asked, and the task enters review. Annotations already mark it non-read-only and non-destructive, and the description substantiates those flags with concrete state changes.
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 concise sentences, with the primary action front-loaded and the routing guidance placed second. Every sentence earns its place; nothing is redundant.
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 simple three-parameter tool with full schema coverage and a clear description, this is nearly complete. It lacks explicit detail on what happens after the human answers, but the review-state mention covers the most important post-invocation 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?
The schema covers all three parameters at 100% with useful descriptions, including the optional recipient's default behavior. The description adds only the 'specific human' notion, which aligns with to_user_id but does not materially extend schema semantics.
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 action sequence (pause, ask, put in review) with a clear resource (specific human, task). Distinguishes from siblings like answer_question by making the direction of communication explicit.
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?
Gives an explicit usage criterion: prefer this when a decision requires human input rather than guessing. Does not name alternative tools, but the guidance is clear enough for an agent to select it appropriately.
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.
Most tools target a distinct resource and action, but several adjacent pairs are easy to confuse: add_comment vs add_progress_note, call_executor vs call_integration, log_client_decision vs update_client_context vs memory_save, and list_context_sources vs list_integrations. The descriptions do disambiguate them, but the boundaries are subtle enough that misselection is likely with 81 tools.
The overwhelming majority follow a clear verb_noun snake_case pattern (create_task, update_project, list_integrations, set_webhook), and get_/ list_/ create_/ update_ families are predictable. Minor deviations exist: memory_read/memory_save/memory_search invert to object_verb, and bare nouns like whoami, glossary, and security_posture break the pattern, but there is no chaotic casing or mixed conventions.
At 81 tools, this is far above the weight that is comfortable for an agent's tool-selection surface, especially since many tools belong to families that could be consolidated (webhooks, worker keys, project logs, context/memory). Although the domain is broad, the count will overwhelm agents and increase misrouting.
The surface is unusually comprehensive: tasks, projects, workers, leases, artifacts, comments, handoffs, webhooks, keys, context, memory, integrations, and support all have create/read/update/delete or equivalent lifecycle coverage. The gaps are minor, such as remove_dependency without a visible add_dependency and no artifact/comment deletion, but agents can work around them.