Skip to content

Security: TCoherence/oh-my-agent

Security

SECURITY.md

Security Policy

Oh My Agent is a single-user, self-hosted tool. The maintainers take security issues seriously and appreciate responsible disclosure.

Reporting a vulnerability

Please do not file a public GitHub issue for a vulnerability report.

Email security reports to tcoherence@gmail.com with:

  • A short description of the issue and its impact.
  • Steps to reproduce, ideally with a proof-of-concept.
  • Affected version (oh-my-agent --version).
  • Your operating environment (local venv vs Docker, Python version, OS).

You should receive an acknowledgement within 72 hours. Expect a substantive follow-up (confirmation, patch ETA, or request for more detail) within 7 days.

Scope

In-scope:

  • The published oh_my_agent Python package
  • Container images built from the repo's Dockerfile and compose.yaml
  • Skill files under skills/ and bundled automation examples
  • Memory / SQLite persistence and the AuthFlow credential storage layer
  • Discord gateway, automation scheduler, and runtime service

Out-of-scope:

  • Vulnerabilities in upstream dependencies (please report to them directly — you can cc us if the issue is triggerable via Oh My Agent)
  • Vulnerabilities that require a compromised host machine or stolen Discord bot token (the threat model assumes the operator protects those)
  • Denial-of-service achievable only by exhausting host resources (disk, RAM, CPU)
  • Social-engineering attacks against the operator

Supported versions

We patch the latest released minor version on the main branch.

Version Supported
0.9.x ✅ (current RC / contract-freeze)
0.8.x ⚠️ security-only, best effort
< 0.8 ❌ no longer supported

Once 1.0 ships, 1.x becomes the supported line and 0.9 moves to security-only.

Credential handling

Oh My Agent stores provider credentials (Discord tokens, CLI auth cookies, etc.) on the local filesystem under ~/.oh-my-agent/auth/. Files are written with chmod 0600. The AuthFlow and credential store only read / write these files; they are never transmitted to external services except as required by the relevant provider (e.g. Bilibili's login API).

If you discover a path where credentials could leak to logs, messages, attachments, or external services, that is in-scope.

Safe harbor

Good-faith security research that follows this policy will not be subject to legal action by the maintainers. Please make a best-effort attempt to avoid privacy violations, data destruction, and service disruption during your testing.

There aren't any published security advisories