describe_tool

Returns the full contract for one catalog tool: its parameter schema, its cost, whether it is healthy right now, and an execute_as template you can copy into execute_tool.

Call describe_tool before every execute_tool. Never guess parameters.

string

required

A canonical Provider/Operation name (Firecrawl/scrape, Brave/getWebSearch), a browse path from find_tools (social/twitter/post_search_tweets), or agentkey_account. On a typo the response contains fuzzy-match suggestions — read them and correct rather than retrying blindly.

Returns

describe_tool("Serper/search")

{
  "name": "Serper/search",
  "provider": "Serper",
  "category": "Search",
  "tags": ["Serper"],
  "summary": "Web search",
  "description": "",
  "cost": { "credits_per_call": 0.2 },
  "health": { "healthy": true },
  "params": {
    "type": "object",
    "required": ["q"],
    "properties": {
      "q":   { "type": "string",  "description": "Search query string." },
      "gl":  { "type": "string",  "default": "us", "description": "Country code (geolocation), e.g. us, cn, gb." },
      "hl":  { "type": "string",  "default": "en", "description": "UI language code, e.g. en, zh-cn." },
      "num": { "type": "integer", "description": "Number of results to return." },
      "tbs": { "type": "string",  "description": "Time-based filter: qdr:h, qdr:d, qdr:w, qdr:m, qdr:y." }
    }
  },
  "execute_as": { "name": "Serper/search", "params": { "q": "<Search query string.>" } }
}

Field	Meaning
`params`	JSON Schema of the parameters: types, `required`, enums, defaults, descriptions. These are the provider’s own parameter names.
`cost.credits_per_call`	Price in credits.
`cost.metered`, `cost.credits_minimum`, `cost.billing_note`	Present on metered tools (for example `Firecrawl/scrape`): the price is a minimum and the actual charge depends on the response. Read `credits_charged` on the execute result.
`health.healthy`	Whether AgentKey’s probes currently see the provider as healthy.
`execute_as`	A template with placeholders for every required parameter. Fill in the values and pass it to `execute_tool`.

Cost guidance

Multiply cost.credits_per_call by your planned call count before executing a batch. For bulk work (three or more calls, or a meaningful credit estimate), check your balance first with the free agentkey_account tool and confirm the plan with the user.

Next step

Copy execute_as, fill in the values, and pass it to execute_tool.

Overview

Scenario Methods

Catalog & Describe

Methods

Tool Catalog (MCP)

Returns

Cost guidance

Next step

​Returns

​Cost guidance

​Next step

Returns

Cost guidance

Next step