Sautikit

create_agent

Create a draft AI voice agent. The response includes tool_signing_secret ONCE — tell the user to store it now; only rotate-secret (dashboard) can produce a new one.

Tool call template

Replace each placeholder with a value matching the input schema. Only required parameters are included; add optional parameters as needed.

{
  "id": 1,
  "jsonrpc": "2.0",
  "method": "tools/call",
  "params": {
    "arguments": {
      "name": "<name>",
      "base_prompt": "<base_prompt>",
      "model": "<model>"
    },
    "name": "create_agent"
  }
}

Input Schema

Name	Required	Description
`name`	Yes	agent display name
`model`	Yes	model id from the workspace's model catalog (GET /v1/models)
`voice`	No	TTS voice id
`accent`	No	speaking accent hint
`purpose`	No	Required on tools that read data or reports; recommended on every tool. One sentence: what you are doing and why (for example "finding this caller's open ticket to answer them"). Building a report, dashboard, analytics or an export is a SautiDesk Pro feature done by the report tools, not by pulling records.
`base_prompt`	Yes	system prompt defining persona and rules
`speech_style`	No	speaking style hint
`max_call_minutes`	No	hard per-call duration cap

Schema Changelog

Changes observed during successful MCP inspections.

First observedNov 5, 2026

TDQS

A3.8/5.0

Behavior4/5

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

Annotations only declare openWorldHint=false and destructiveHint=false, so they add little. The description supplies the key non-obvious behavior: the created agent is a draft, and the tool_signing_secret is returned exactly once with no API path to regenerate it. It omits permissions/scope requirements and whether the draft is immediately usable.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Two dense sentences, front-loaded with the action and then the one fact an agent will otherwise get wrong (single-shot secret). No filler or restated title.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

There is no output schema, and the description correctly compensates by naming the one consequential return field. For a create-mutation with fully documented parameters this is nearly complete; only the lifecycle between draft and publish_agent is left implicit.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters3/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, with all 8 parameters documented inline including the model-catalog lookup and the purpose field. The description adds no parameter-level meaning beyond the schema, so the baseline 3 applies.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose4/5

Does the description clearly state what the tool does and how it differs from similar tools?

States a specific verb and resource ('Create a draft AI voice agent'), and the word 'draft' distinguishes it from publish_agent and update_agent without naming them. An agent can tell this is the creation entry point rather than a modification or listing tool.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines3/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

Usage is implied by the verb: create a new agent. There is no explicit statement of when to prefer update_agent over this tool, nor of prerequisites such as workspace-scoped model catalog access. The only explicit guidance is operational: store the returned secret immediately.

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

All Sautikit tools Inspect

Try in Browser

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources

Need Help?