Dev Server
Iterate on your agents in seconds. connic dev syncs your local files to a cloud runner with hot-reload. Edit, save, run, repeat.
Overview
Cloud-backed development loop without git commits or deployments
The connic dev command spins up an isolated dev runner in the cloud, uploads your local project, and watches your filesystem for changes. Edits land in the runner within 2–5 seconds and are picked up by the agent loader on the next invocation. You can:
- Iterate in seconds: save a file, trigger your agent, see the result
- Use real connectors: wire your dev environment to webhooks, cron jobs, Telegram, etc., the same way production does
- Skip git entirely: no branches, no commits. Your editor is the source of truth
- Stay isolated from production: dev runs in its own environment with its own API keys, env vars, and connectors
• Connic SDK installed: pip install connic-composer-sdk
• A valid agent project with at least one agent in agents/
Quick Start
Authenticate
Run connic login in your project root. The CLI opens the dashboard for you to create an API key, then prompts for the login token to paste back.
connic loginCredentials are saved to .connic. Add the file to your .gitignore.
Start a Dev Session
Start a session to provision a runner and upload your files.
# Start an ephemeral dev session (auto-deleted on exit)
connic devThe CLI uploads your project, boots the runner container, and starts watching your filesystem.
Develop and Iterate
Edit your agent YAMLs, tools, middleware, or tool hooks. The CLI uploads on each save (debounced ~1s). Use the dashboard's connectors page or the agent detail page to trigger your agents and inspect runs in real time.
Press Ctrl+C to stop the session.
Ephemeral vs Named Environments
Dev sessions can be ephemeral or named. The choice changes whether the underlying environment, its connectors, and its env vars persist after you exit.
connic dev- • Auto-deleted on Ctrl+C
- • Fresh environment every run
- • Best for quick experiments and bug repros
# Reuse a named dev environment across sessions
connic dev my-feature- • Persists after Ctrl+C
- • Reattach later with the same name
- • Best for ongoing feature work where connectors and env vars matter
File Watching
The CLI watches the standard project directories for changes:
Saves are debounced for 1 second so a multi-file rename or formatter pass uploads as one batch. Subfolders inside agents/ and tools/ are watched recursively, so larger projects can split files into namespaces without changing anything.
Tip: Adding a new package to requirements.txt triggers a re-install in the runner on the next sync. No restart needed.
Dashboard Integration
Dev environments appear alongside your standard environments:
- Environment switcher: dev environments are grouped separately so you can't accidentally point a webhook at production
- Agents view: shows the agents currently loaded by your dev runner
- Connectors: create and trigger them in the dev environment exactly like production
- Deployments page: the dev session shows up with a Dev session badge so you can spot it in the list
Configuration
Authentication via file or environment variables
The .connic file in your project root stores your credentials:
{
"api_key": "cnc_xxxxxxxxxxxx",
"project_id": "your-project-uuid"
}You can also pass them through environment variables, useful for CI or shared shells where a checked-in file would be wrong:
# Environment variables
CONNIC_API_KEY=cnc_xxxxxxxxxxxx
CONNIC_PROJECT_ID=your-project-uuidConflict Handling
Only one dev session can be active per named environment at a time. Starting a second session against the same name returns:
Error: A dev session is already active for this environment. Stop the existing session first with Ctrl+C.
Use a different name, or omit it for an ephemeral session, to work in parallel.
With connic dev you get the cloud's connectors and runtime with the latency of a local edit-save loop.