Prinevo documentation
Build with Prinevo
Configure your agent team, choose the right delivery path, run work in cloud sandboxes, and review evidence before changes ship.
Start here
Recommended first guide How to use Project Lane Take a meaningful feature from request through planning, implementation, review, verification, and learning. Open the guide
Choose a laneUnderstand when to use Full lane, Quick lane, Plan, or Direct agent.
Configure agentsChoose a model, connection, reasoning level, and compute profile for each agent.
Track usageReview tokens, sessions, duration, and cost by user, agent, model, task, or session.
Add project skillsGive agents additive company playbooks without weakening Prinevo safety baselines.
Core concepts
- Session
- One durable conversation and its agent work, decisions, artifacts, and evidence.
- Lane
- The execution path for a request. Full lane coordinates the complete workflow. Quick lane handles bounded work. Direct agent runs one specialist.
- Agent
- A specialist with its own role, model, connection, reasoning level, compute profile, skills, and tool permissions.
- Cloud workspace
- An isolated compute environment where an agent can inspect repositories, build changes, and run verification.
- Evidence
- Artifacts such as plans, diffs, tests, logs, screenshots, and verification results that travel with the work.
Before your first run
Set up the execution choices once, then every new run starts with an explicit repository, model, connection, reasoning, and compute configuration.
- 1Connect source control
Add your provider and select the repositories Prinevo may prepare.
- 2Check Factory
Confirm repository setup and analysis are ready for agent work.
- 3Configure Agents
Save the model, connection, reasoning, and compute choices that new runs should snapshot.
- 4Start a session
Choose a lane based on the size and risk of the work.