Developer API
Build on Cuneiform Chat from your own code
A public, key-authenticated REST API that lets your code do what the dashboard does — feed knowledge in, manage your AI agents, and query them from anywhere. One bearer key, predictable JSON, and a reference you can copy from.
What you can build with the API
Ingest knowledge programmatically
Push documents into an agent's knowledge base straight from your code or pipeline. Files are parsed, chunked, embedded, and made searchable — the same RAG stack the dashboard uses, driven by an API call.
Manage agents as code
Create, list, update, and inspect your AI agents through the API. Wire agent configuration into your own provisioning or CI so a new agent is a deploy step, not a dashboard click.
Query an agent from anywhere
Send a question to any agent and get a grounded answer back, with the same retrieval and reasoning your channels use. Embed it in an app, a backend job, or another product.
Scoped, revocable API keys
Mint keys with only the scopes a workload needs and revoke them the moment they leak. Keys are prefixed cuk_live_ and shown once — least privilege, by design.
Predictable errors and rate limits
Every response is JSON with a stable shape, consistent error codes, and clear rate-limit headers — so your client can retry, back off, and surface failures without guessing.
A reference you can copy from
A full published reference covers authentication, conventions, errors, rate limits, and every endpoint across all three domains — with request and response examples ready to paste.
From key to first response
Five steps from an empty terminal to a live integration.
Get a key
Open the developer console in your dashboard and issue an API key, choosing the scopes the integration needs. Copy it once — it starts with cuk_live_.
Authenticate
Send the key as Authorization: Bearer on every request (the X-API-Key header is a documented alias). No OAuth dance, no session to refresh.
Call an endpoint
Hit any path under /api/developer/v1 — ingest into Knowledge, manage Agents, or run an Agent Query. Same base URL, same auth, three domains.
Handle the response
Parse the JSON body, read the rate-limit headers, and branch on the error code when something goes wrong. The shape is stable across every endpoint.
Ship and scale
Move from a curl test to production: rotate keys, scope per workload, and let the documented rate limits guide your client's retry and concurrency.
Use Cases
What teams build on the API
Sync knowledge from your systems
Wire a nightly job or a webhook to push new docs, help-center articles, or product specs into an agent's knowledge base — so answers stay current without anyone touching the dashboard.
Embed an agent in your own product
Call the query endpoint from your app or backend to add a grounded AI assistant to a surface we don't ship a channel for — your mobile app, an internal tool, a custom widget.
Provision agents in CI/CD
Create and configure agents as part of a deploy pipeline, so spinning up a new environment or customer tenant includes its AI agent as code, not a manual setup step.
Automate back-office workflows
Have a script ask an agent to summarize, classify, or answer over your knowledge base on a schedule, and route the result wherever your team already works.
Power internal tools and dashboards
Pull agent answers into a Slack bot, an ops dashboard, or a support console you own — one bearer key and a JSON response is all the integration needs.