Skip to content

Security: Diplomat-ai/diplomat-agent

Security

SECURITY.md

Security Policy (v0.2.0)

Reporting a vulnerability

Do not open a public GitHub issue for security vulnerabilities.

Email: josselin@diplomat.run Response time: 48 hours

Scope

diplomat-agent runs entirely locally. It makes no network calls and collects no telemetry. It only reads Python files — it does not execute them.

Vulnerabilities in scope:

  • Malicious Python files that cause unexpected behavior during AST parsing
  • Patterns that create false negatives on critical tool calls (e.g. payment functions not detected)
  • Bypass of --fail-on-unchecked CI gate

Out of scope

  • Vulnerabilities in scanned codebases (diplomat-agent reports them, it doesn't fix them)
  • Feature requests (use GitHub issues)

There aren't any published security advisories