share-usage
Fail
Audited by Gen Agent Trust Hub on Jun 18, 2026
Risk Level: HIGHDATA_EXFILTRATIONCOMMAND_EXECUTION
Full Analysis
- [DATA_EXFILTRATION]: The skill reads highly sensitive Claude Code interaction logs from ~/.claude/history.jsonl and the ~/.claude/debug/ directory. These files contain full conversation histories, tool results, and other internal agent data. It then transmits derived content to an external GitHub repository via the gh tool. Although the skill claims to anonymize data, the automated extraction and public sharing of derived log data creates a significant privacy and security risk.
- [COMMAND_EXECUTION]: The skill is vulnerable to command injection through the unsafe construction of shell commands. It interpolates user-provided text (research domain, comments) and data parsed from local logs into a 'gh api graphql' command string without proper escaping or sanitization. This allows for the execution of arbitrary commands with the user's privileges.
- [COMMAND_EXECUTION]: The skill exhibits an indirect command injection surface by processing untrusted log data.
- Ingestion points: ~/.claude/history.jsonl and debug logs.
- Boundary markers: None.
- Capability inventory: Local file writing, shell command execution via gh CLI.
- Sanitization: Absent; the skill performs simple string interpolation of log-derived content and user comments into a shell command query.
Recommendations
- AI detected serious security threats
Audit Metadata