Choose what an Cannon AI agent may do on its own
Set an agent to read only, low-risk actions, or high-risk actions that need approval.
You choose this when you connect an agent on Cannon AI Agents (under Company).
- Press Connect an agent and give it a Name.
- Under What it may access, tick the areas it can reach. Its tools only work inside those areas.
- Optionally tick specific Tools. Leave them all unchecked to allow every tool its access permits.
- Under May act on its own up to, pick:
- Under Approvals go to, pick Push notification, Team Chat or Email (then type the Approval email).
- Press Create connection.
- Read only: it can look but can't change anything. - Low-risk actions: it can also do simple things. - High-risk actions (with approval): it can propose risky actions, but each one waits for a person to approve it.
The agent's token shows once: copy it into your assistant's connection settings, then press Done. Each connected agent shows its limit, such as "up to: Read".
Was this helpful?
