Docs for AI agents
These docs are available in forms that an AI agent can load without a browser or a key: plain-text indexes, one Markdown file per page, and an OpenAPI spec. Point your agent at them instead of a rendered page, because Markdown is far smaller than HTML and costs far fewer tokens.
llms.txt and llms-full.txt
Section titled “llms.txt and llms-full.txt”/llms.txt is a short index. It has instructions for agents, a link to every docs page and its Markdown version, and the list of API endpoints. /llms-full.txt is the whole documentation in one file, for an agent that wants everything in a single read.
curl https://kenari.id/llms.txtcurl https://kenari.id/llms-full.txtPages as Markdown
Section titled “Pages as Markdown”Add .md to the URL of any docs page to get clean Markdown, for example https://kenari.id/en/docs/chat.md. https://kenari.id/docs.md holds every page in one Markdown file.
curl https://kenari.id/en/docs/chat.mdcurl https://kenari.id/docs.mdThe Markdown files and llms-full.txt are generated from the English pages. The Indonesian URLs, such as https://kenari.id/docs/chat.md, serve the same English text.
OpenAPI
Section titled “OpenAPI”https://kenari.id/openapi.json is the OpenAPI 3.1 spec for the /v1 API. It lists every /v1 endpoint with its request and response shapes, the error format, and the bearer scheme for kn- keys. Use it with code generators, Postman and API explorers.
curl https://kenari.id/openapi.jsonLive model list
Section titled “Live model list”Do not copy model ids or prices from a page. GET /v1/models is public and returns the current models with their prices in Rupiah. See Models and pricing.
curl https://kenari.id/v1/modelsMCP server
Section titled “MCP server”An agent that speaks MCP can search these docs, read the model list and call search tools through https://kenari.id/mcp. See MCP server for setup and the tool list.
Instructions to give an agent
Section titled “Instructions to give an agent”Paste this block into the system prompt, AGENTS.md or CLAUDE.md of any agent that will use kenari.
You are working with the kenari API (https://kenari.id), an LLM gateway billed in Rupiah.- Docs index: https://kenari.id/llms.txt. Full reference in one file: https://kenari.id/llms-full.txt.- Add .md to a docs page URL to read it as Markdown, for example https://kenari.id/en/docs/chat.md.- API spec: https://kenari.id/openapi.json.- Live model list and prices: GET https://kenari.id/v1/models. Never hardcode model ids or prices.- OpenAI-compatible clients use the base URL https://kenari.id/v1.- Anthropic SDKs and Claude Code use the base URL https://kenari.id, without /v1.- Authenticate with the header Authorization: Bearer $KENARI_API_KEY. Never print or commit the key.- Test with a :free model id first.- MCP server: https://kenari.id/mcp.