Change your agent in the editor.
Try it on real messages seconds later.
Every save runs in your own cloud environment, with its own channels and data. You try changes on real chats and webhooks before anything reaches your customers.
Test the whole flow in the channels your customers use.
A dev environment has its own connections, so chats, schedules, webhooks and your app reach the code you just saved. Production keeps running as it is.
In the dev environment, a schedule starts the reminders and your own test bot sends them. Adjust the wording until it reads right, before a member gets one.
The member system's test setup sends cancellations straight to the dev environment, and each one runs your latest code. The agent's win-back draft waits for a person to send it.
connic builds the members' schedule app on your machine and serves it from the dev environment. Save a page or a prompt, reload, and it talks to your latest agent.
Try anything. Your customers never notice.
Each dev session runs in its own environment. Pick it in the dashboard to start its agents, open its runs and set up its connections.
- Test data stays out of production
- Its own variables, connections, database, document search and sessions.
- Throw it away or keep it
- A quick test is deleted when you quit. A named environment keeps its setup for the next session.
- The same models as production
- Model access belongs to the project, so connic models and your own provider keys work without extra setup.
One command from your editor to a running environment.
connic dev uploads the project, then every save. connic checks the agents first, so a broken save prints the reason and the last good version keeps running.
- What it watches
agents/,tools/,middleware/,hooks/,schemas/,guardrails/,tests/,connectors/andapps/. Connector and app files apply to the dev environment like a deploy.- Keys while it runs
ruploads right away,truns your test cases against the files you just saved, andqends the session and cleans up.- New dependencies
requirements.txtinstalls when the session starts. After you change it, start the session again.
Keep exploring
- composer SDKWrite agents in YAML and tools in code, and see each save in the next run.
- TestingTurn what you tried by hand into cases that run before every deploy.
- EnvironmentsGive staging and production their own variables and connectors too.
- ObservabilityOpen the trace of every dev run, tool call by tool call.