| Version | Supported |
|---|---|
| 2.x | ✅ Active |
| 1.x | |
| < 1.0 | ❌ End of life |
Do not open a public issue for security vulnerabilities.
Email fullerbt@users.noreply.github.com with:
- Description of the vulnerability
- Steps to reproduce
- Affected version(s)
- Potential impact assessment
| Stage | Timeline |
|---|---|
| Acknowledgment | Within 48 hours |
| Initial assessment | Within 5 business days |
| Fix for Critical/High | Within 7 days |
| Fix for Medium/Low | Within 30 days |
| Public disclosure | After fix is released |
| Severity | Description | Examples |
|---|---|---|
| Critical | Remote code execution, credential theft | Prompt injection that exfiltrates secrets |
| High | Privilege escalation, data exposure | Hook script that bypasses safety guards |
| Medium | Information disclosure, denial of service | Unsafe command patterns in skill templates |
| Low | Minor information leak, best practice violation | Verbose error messages revealing paths |
This repository contains Claude Code plugins — text-based configuration files (skills, agents, hooks) that instruct Claude how to behave, plus a Go CLI (ao).
- Prompt injection vulnerabilities in skill/agent definitions
- Unsafe bash commands in hook scripts
- Credential exposure in examples or templates
- Command injection in the
aoCLI - Dependency vulnerabilities in Go modules
- Unsafe file operations in scripts
- Claude Code CLI vulnerabilities → report to Anthropic
- General Claude model behavior → report to Anthropic
- Social engineering attacks
- Vulnerabilities requiring physical access
We consider security research conducted in good faith to be authorized. We will not pursue legal action against researchers who:
- Make a good faith effort to avoid privacy violations, data destruction, and service disruption
- Report vulnerabilities promptly and provide sufficient detail to reproduce
- Do not exploit vulnerabilities beyond what is necessary to demonstrate the issue
- Do not publicly disclose vulnerabilities before a fix is available
- Reporter submits vulnerability via email
- Maintainer acknowledges receipt within 48 hours
- Maintainer assesses severity and confirms timeline
- Maintainer develops and tests fix
- Maintainer releases fix and publishes advisory
- Reporter credited in release notes (unless anonymity requested)
We gratefully acknowledge security researchers who help keep AgentOps safe. Contributors will be credited in release notes unless they prefer to remain anonymous.