Safety And Governance
Agent safety in Veoveo starts with workspace boundaries, bot identity, permissions, tool availability, and reviewable operation history.
Governance Principles
- A bot should act only inside the workspace and channels configured for it.
- Tool access should match the bot's role and the team's workflow.
- Durable changes should be visible through artifacts, links, events, or operation history.
- People should be able to review source context before trusting an output.
- Sensitive information should stay out of prompts, shared links, and public examples unless the workspace is intended to contain it.
Admin Review Checklist
Before enabling or expanding a bot workflow, review:
- who can reach the bot
- which channels are connected
- which tools are available
- what artifacts or outputs the bot can create
- how failed operations are surfaced
- how users should verify results
Common Risks
- Broad prompts that do not name the artifact or workspace context.
- Tools enabled for a bot that does not need them.
- Channel bindings that reach the wrong audience.
- Treating generated summaries as source evidence.
- Missing review of operation timelines after high-impact work.
Media To Add
- Table: governance checklist with bot identity, channel, tool, output, and review owner columns. It helps admins review bot setup. Source: workspace bot configuration process.
- Diagram: permission boundary from user to channel to bot to tool to artifact output. It helps explain where control is enforced. Source: bot settings and tool settings pages.