API Reference - Hexel Studio

The Hexel Studio API provides programmatic access to all platform capabilities. Every feature in the docs and Console is backed by this API.

Base URLs

All API calls (except agent traffic) go through a single gateway host. Deployed-agent and sandbox traffic use separate subdomain endpoints.

Surface	Base URL
Compute, Registry, Orchestration, Tools, Data	`https://api.hexelstudio.com`
Token service (STS)	`https://sts.hexelstudio.com`
Running agent traffic	`https://agent-<name>-<hash>.compute.hexelstudio.com`
Running sandbox traffic	`https://sandbox-<id>.compute.hexelstudio.com`

Path prefixes

Product	Prefix
Compute	`/compute/v1/`
Agent Registry	`/registry/v1/`
Orchestration (tasks, fleets)	`/orchestrator/v1/`
Approvals	`/v1/approvals/`
Policies	`/v1/policies/`
Tool Gateway	`/tools/v1/`
Data Platform	`/data/v1/`

Authentication

Every request requires a Bearer token obtained from the STS. Exchange your API key or OAuth client credentials for a short-lived token (valid 15 minutes):

# API key exchange
TOKEN=$(curl -s -X POST https://sts.hexelstudio.com/token \
  -H "X-API-Key: YOUR_API_KEY" | jq -r .access_token)

# Use the token
curl https://api.hexelstudio.com/compute/v1/instances \
  -H "Authorization: Bearer $TOKEN"

The SDKs handle token exchange and refresh automatically. See Authentication for OAuth client credentials and environment variable configuration.

Error responses

Each service returns a consistent error body, but the exact shape varies by product:

Product	Error shape
Compute	`{ "error": "code", "message": "..." }`
Data Platform	`{ "error": { "code": "DOMAIN.CODE", "message": "...", "request_id": "...", "details": {} } }`
Tool Gateway	`{ "error": { "code": "CODE", "message": "..." } }`
Orchestration (rate limit)	`{ "error": "rate limit exceeded", "limit": N, "window_seconds": N }`

The Data Platform error shape includes request_id. Include it when contacting support — it ties directly to the failing request in platform logs.

Rate limits

Product	Limit
Data Platform	10,000 requests / organization / hour
Orchestration	100 requests / organization / minute
Compute	Per-organization quota (see Tiers & Limits)

Exceeding a limit returns 429. The response includes X-RateLimit-Limit, X-RateLimit-Remaining, and Retry-After headers where applicable.

Browse the reference

Use the sidebar to browse each endpoint group. Every endpoint has an interactive playground where you can send real requests and inspect responses.

Group	Description
Sandboxes	Allocate, execute, renew, release isolated environments
Agents	Register and manage agent images in the registry
Instances	Deploy agents to permanent endpoints
Skills	Publish and manage reusable capabilities
Usage	Query compute consumption by tier and time range
Access Clients	Manage OAuth clients for M2M access

​Base URLs

​Path prefixes

​Authentication

​Error responses

​Rate limits

​Browse the reference

Base URLs

Path prefixes

Authentication

Error responses

Rate limits

Browse the reference