For agents
How an AI agent finds, calls and pays for a Tenhatch tool.
Tenhatch builds every tool for machines as well as for people. Start at one index for all of them, then follow each tool’s own documents. Nothing here needs a browser.
Start at the index
Both files are generated from the same list as this site, so they never disagree with it. Each tool carries a status; links and endpoints appear when a tool is live.
tenhatch.com/llms.txtEvery tool in plain text
What each tool does, who it is for, its prices, how agents call it and its status.
tenhatch.com/.well-known/tenhatch.jsonEvery tool as JSON
Slug, status, kind, prices for people and for agents, the agent door and, once live, its endpoints.
What every web tool exposes
Each web tool lives on its own .com and serves the same four documents at the same paths, so an agent that has used one knows where to look on the next.
/llms.txtA summary written for language models
What the tool does, its inputs and outputs, prices, endpoints and policies, in plain text.
/.well-known/product.jsonA structured description
The same facts as JSON: the feature, its endpoint, authentication, plans and the price per result.
/api/v1/openapi.jsonAn OpenAPI 3.1 document
Generated from the schemas the server validates with, so the description and the API agree.
/docs/agent-accessAn MCP server
The tool’s feature as MCP tools, with the same input schema as the API. Setup is on this page of each tool.
How agents pay
No seats. Agents pay for results, through an account a person sets up once, and see the price before they call.
Authorization: Bearer <api key> X-Usage-Source: agent
A person creates the account and the API key on the tool’s site. Calls use the plan’s monthly allowance and are billed per result after it, at the price in that tool’s llms.txt. The usage header only labels the call; it never changes the price.
Apify API or Apify’s MCP server
The data tools are Apify Actors. Run them from an Apify account and pay per result on Apify; the price per result is on each Actor’s Apify Store page.
For people only
InStock First, FeedTrim, OrderColor run inside Shopify and Chrome and have no agent door.
Tool by tool
| Tool | How agents call it | Agent price | Status |
|---|---|---|---|
| AccessDiff | API, OpenAPI, MCP, llms.txt | Pay per scan | Early access |
| CheckoutBeat | API, OpenAPI, MCP, llms.txt | $0.25 per check | Early access |
| Career Site Jobs | Apify API or Apify's MCP server | Pay per result | Early access |
| App Store Watch | Apify API or Apify's MCP server | Pay per result | Early access |
| Storefront Catalog | Apify API or Apify's MCP server | Pay per result | Early access |
Good manners
- Respect
Retry-Afteron HTTP 429. On web tools, limits are published in theX-RateLimit-*response headers. - Send
X-Usage-Source: agenton web tools so usage is labelled as yours. - Read a tool’s llms.txt before calling it; prices and limits are stated there.
- The data tools read public business data only. Do not use them to collect data about people.
Questions from an agent or its owner: support@tenhatch.com. Back to the tools.