AgentShield is a security-focused project. If you discover a security vulnerability, please report it privately so it can be handled before public disclosure.

Security fixes are prioritized for the latest released version and the main branch. Older releases may receive fixes when the issue is severe and the patch can be applied safely.

Please do not disclose security vulnerabilities in public issues, discussions, or pull requests before maintainers have had a chance to investigate.

Preferred reporting path:

1. Use GitHub private vulnerability reporting or a GitHub Security Advisory for this repository, if available.
2. If private reporting is not available, open a minimal public issue asking for a secure contact path, without including exploit details, secrets, or proof-of-concept payloads.

Include, when safe to share privately:

• affected component and version, or commit SHA;
• attack scenario and impact;
• reproduction steps and payload examples;
• scanner command or integration path used;
• files, logs, or commands that demonstrate the issue;
• whether credentials, source code, or runtime data could be exposed.

In scope:

• vulnerabilities in AgentShield detection, output, CLI, GitHub Action, runtime guard, or VS Code extension behavior;
• unsafe handling of findings, SARIF, HTML, JSON, logs, or runtime events;
• secret leakage, command injection, path traversal, unsafe archive handling, or sandbox bypasses caused by AgentShield itself.

Out of scope:

• vulnerabilities in third-party projects scanned by AgentShield;
• intentionally vulnerable test fixtures under tests/fixtures/;
• placeholder tokens used by redaction tests;
• findings that require malicious local filesystem access outside AgentShield's threat model.

• We aim to acknowledge reports quickly and will keep you informed of triage and fix status.
• We may request additional details or a proof-of-concept patch.
• Once fixed, we will publish disclosure details in release notes or an advisory when appropriate.