Features

These pages cover how the Pathvela agent behaves — what it remembers between sessions, how it asks before doing anything risky, and how to get the best work out of it. They apply in both the CLI and the SDK.

For what you set up to shape the agent — rules, skills, tools, MCP — see Configuration.

In this section

• Memory — how the agent remembers useful facts across sessions, and how to toggle it.
• Permissions & safety — how approvals work, read-only vs full access, and the guardrails on risky commands.
• Best practices — how to work with the agent so it does its best work.