Developer documentation

Build on the
Clymora platform.

API references, SDKs, and implementation guides for building, deploying, and governing AI in production.

Authorized access only · Sessions are encrypted
Back to site Secure access

Sign in to continue

Enter your Clymora credentials to open the documentation portal.

Incorrect email or password. Please try again.

Trouble signing in? Contact your workspace administrator.

AK
Aayush Kumar
aayushkumar@clymora.com
Documentation / Overview

Overview

Welcome to the Clymora developer documentation. The Clymora API lets you deploy domain-tuned models, orchestrate AI agents, and stream real-time inference — all governed inside your own perimeter.

AK
Welcome back, Aayush Kumar
Developer · Clymora Workspace · Full access
Verified

What you can build

Clymora is a unified platform for shipping AI into production. Whether you are wiring a single completion endpoint into an app or orchestrating fleets of autonomous agents across your enterprise systems, the same API surface, governance model, and observability tools apply.

  • Inference — run completions against Clymora foundation models or your own fine-tunes.
  • Agents — execute multi-step agent workflows that reason, plan, and call tools.
  • Governance — audit trails, policy enforcement, and explainability on every request.
  • Streaming — token-level streaming for real-time, low-latency experiences.

Jump in

Documentation / Quickstart

Quickstart

Make your first authenticated request in three steps.

01

Generate an API key

Create a secret key from the Authentication section. Keep it server-side.

02

Install the SDK

Add the official client for your language of choice — see SDKs & libraries.

03

Send a request

Call the inference endpoint and stream your first completion.

Your first request

Set your key as an environment variable, then call POST /v1/inference:

curl https://api.clymora.com/v1/inference \
  -H "Authorization: Bearer $CLYMORA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "clymora-core",
    "input": "Summarize Q3 results."
  }'
from clymora import Clymora

client = Clymora(api_key=os.environ["CLYMORA_API_KEY"])

resp = client.inference.create(
    model="clymora-core",
    input="Summarize Q3 results.",
)
print(resp.output)
import Clymora from "clymora";

const client = new Clymora({ apiKey: process.env.CLYMORA_API_KEY });

const resp = await client.inference.create({
  model: "clymora-core",
  input: "Summarize Q3 results.",
});
console.log(resp.output);

Tip: add "stream": true to receive tokens as they are generated over server-sent events.

Documentation / Authentication

Authentication

All requests are authenticated with a bearer token. Keep your secret key server-side and never expose it in client code.

Your secret key

Every workspace is issued a live secret key. Include it as a bearer token on every request.

cly_live_9f2c7a••••••••4f8a
Authorization header
Authorization: Bearer cly_live_9f2c7a...4f8a

Never expose your secret key in browser code, mobile apps, or public repositories. Rotate compromised keys immediately from the workspace console.

Core endpoints

The most common endpoints you will authenticate against:

MethodEndpointDescription
POST/v1/inferenceRun a model completion.
POST/v1/agents/runExecute an agent workflow.
GET/v1/modelsList available models.
PUT/v1/models/{id}/tuneFine-tune on your data.
Documentation / API reference

API reference

The Clymora REST API is organized around resources. All endpoints are served over HTTPS from https://api.clymora.com and return JSON.

POST  /v1/inference

Run a completion against a Clymora model or one of your fine-tunes.

ParameterTypeDescription
modelstringRequired. Model ID, e.g. clymora-core.
inputstringRequired. The prompt or input text.
streambooleanStream tokens over SSE. Defaults to false.
max_tokensintegerMaximum tokens to generate. Defaults to 1024.
200 · application/json
{
  "id": "inf_9f2c7a41",
  "model": "clymora-core",
  "output": "Q3 revenue rose 18% to $4.2B...",
  "usage": { "input_tokens": 12, "output_tokens": 86 }
}

POST  /v1/agents/run

Execute an agent workflow. Agents can reason across multiple steps and call registered tools before returning a final result.

ParameterTypeDescription
agentstringRequired. The agent ID to run.
inputobjectRequired. Structured input for the workflow.
toolsarrayTool definitions the agent may call.

GET  /v1/models

List every model available to your workspace, including base models and your fine-tunes.

PUT  /v1/models/{id}/tune

Start a fine-tuning job on your own dataset. Returns a job handle you can poll or subscribe to via webhooks.

Documentation / SDKs & libraries

SDKs & libraries

Official clients for Python, Node, Go, and Java — with typed responses, automatic retries, and streaming helpers built in.

Install

pip install clymora
npm install clymora
go get github.com/clymora/clymora-go
<dependency>
  <groupId>com.clymora</groupId>
  <artifactId>clymora-java</artifactId>
  <version>1.4.0</version>
</dependency>

Language clients

Documentation / Webhooks

Webhooks

Subscribe to inference and training events so your systems react the moment work completes — no polling required.

How it works

Register an HTTPS endpoint in your workspace console and choose the events to subscribe to. Clymora delivers a signed JSON payload to your URL for every matching event, with automatic retries and exponential backoff.

Event types

  • inference.completed — a completion finished successfully.
  • inference.failed — a completion errored or timed out.
  • training.started — a fine-tuning job began.
  • training.completed — a fine-tune finished and is ready to deploy.

Example payload

POST · your endpoint
{
  "event": "training.completed",
  "created": 1730583000,
  "data": {
    "model_id": "ft_clymora_core_a1b2",
    "status": "ready"
  }
}

Verify signatures. Every request includes a Clymora-Signature header. Validate it against your signing secret to confirm authenticity before processing.

Documentation / Guides

Guides

Practical, end-to-end walkthroughs for the most common Clymora implementation patterns.

Documentation / Changelog

Changelog

Everything new on the Clymora platform. Subscribe via the webhooks API to get notified programmatically.

v1.4.0
Jun 2025
New

Agent workflows are now generally available via POST /v1/agents/run, with tool calling and multi-step reasoning.

v1.3.2
May 2025
Improved

Streaming latency reduced by 40% at p99. The Python and Node SDKs ship new async streaming helpers.

v1.3.0
Apr 2025
New

Fine-tuning API (PUT /v1/models/{id}/tune) opened to all workspaces, plus a Go client.

v1.2.1
Mar 2025
Fixed

Resolved a rare token-accounting mismatch on very long streamed completions.

Documentation / Support

Support

Reach the Clymora engineering team directly. We aim to respond to workspace issues within one business day.

© 2025 Clymora Inc. · Documentation portal · Signed in as aayushkumar@clymora.com