Setup wizard
Choose your tracker, your runner, and a workflow template. The wizard generates the install command, a .conduit/workflow.md template, the matching .env, and the commands to run — all customized for your selection. Edit the inline fields (repo, team key, model, etc.) and the snippets update live.
The three workflow templates cover the common ways teams use Conduit:
- Autonomous — the agent runs to completion, posts comments, and transitions the issue through every lifecycle event.
- Human-in-the-loop — the agent investigates and prepares the change, then transitions the issue to a review state for a human to take over.
- Customer reply (non-coding) — the agent drafts a reply for a human to review and send. No code changes, no transitions.
-
1. Pick a tracker
-
2. Pick a runner
Tune for my hardware
Throughput depends on your exact setup — model size, quantisation, context length, OS, drivers, and concurrent load. Try a smaller model first if you're unsure.
-
3. Pick a workflow template
Agent runs to completion: comments and transitions the issue through every lifecycle event.
Agent investigates and prepares the change, then transitions the issue to a review state for a human to take over.
Agent drafts a reply for a human to review and send. No code changes, no state transitions.
-
4. Commands
What next
Section titled “What next”- Quickstart — full walkthrough of a Linear + OpenAI setup.
- Configuration — workspace and state paths, secrets, plugins.
- Tracker writes — enable comments and state transitions when you’re ready.
Reference workflow files matching the three templates above: