https://response.stateset.com/api/v1.
Looking for orders, inventory or payments? Start with the embedded commerce engine.
For other services, use the API directory.
Before you start
You need:- A ResponseCX workspace and an API key from Settings → API keys, with
agents:readandagents:write. - A terminal with Bash,
curl7.76 or later, andjqfor reading JSON. - The same terminal session for the commands below so environment variables remain available.
--fail-with-body so an HTTP error exits unsuccessfully while preserving the
server’s explanation.
Step 1 — Check your key
List the agents your key can access:data array. A workspace with existing agents returns those agents.
This is an abbreviated empty response:
200 and a JSON list. Resolve a 401 or 403 using
troubleshooting before continuing.
Step 2 — Create an inactive agent
Choose a key for this creation attempt. Keep the same key and request body if you retry it; choose a new key when you intentionally create another agent.201 with an id and activated: false.
The response is saved in agent.json. It includes fields such as:
agent.json for the error and stop here. Do not substitute the
illustrative ID above. The create endpoint documents
the full request, response and idempotency errors.
Step 3 — Read back what you created
AGENT_ID, the instructions match your request, and
activated is false. You have created a configuration; this walkthrough has not yet
connected a customer channel or sent a customer message.
Ask the workspace what to configure next:
agents for your new agent and next_action for the workspace’s suggested next step.
The workspace endpoint also reports rule, function
and eval counts for each agent.
Step 4 — Activate when ready
Use the links below to add knowledge, tools, rules and evals appropriate to your use case. When the configuration is ready, activate your agent:200 with the same agent ID and activated: true. Activation changes the
agent’s state; use the ResponseCX platform guide
to continue with channel setup and operation.
Deactivate your quickstart agent
When you finish experimenting, setactivated back to false. This keeps its configuration
available for further testing:
Next steps
Add knowledge
Give your agent FAQs, policies and product information.
Connect a tool
Configure a function for an operation such as looking up an order.
Add rules
Define the behavior and escalation rules for your use case.
Create an eval
Record expected behavior so you can check it after configuration changes.
Troubleshooting
Check whether the key exists without printing its value: