Trust model

Security

Aftermath executes repository commands to produce verification evidence. Treat it as a privileged local tool.

• No shell interpolation for planned commands (argv spawn)
• Destructive patterns require approval
• Strict config schema validation for .aftermath.toml
• Best-effort secret redaction and log size limits
• No required cloud backend or API keys

Do not run against untrusted repositories without reviewing .aftermath.toml, package scripts, and CI workflows that may be discovered.

Full policy: SECURITY.md · threat model