claude-code-plugin-troubleshoot

Warn

Audited by Gen Agent Trust Hub on Jul 10, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONDATA_EXFILTRATIONREMOTE_CODE_EXECUTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The skill employs eval in Phase 4 to execute command strings dynamically parsed from hooks.json files within the plugin cache. This pattern allows the execution of arbitrary code stored in local configuration files that the skill is intended to audit.
  • [DATA_EXFILTRATION]: The skill performs broad read operations on AI agent settings (~/.claude/settings.json) and scans the plugin cache directory. While relevant for troubleshooting, these files may contain sensitive configuration metadata or environment details.
  • [COMMAND_EXECUTION]: The skill executes bulk file system modifications, including using sed to rewrite shell script logic and chmod +x to change execution permissions for files located in the plugin cache.
  • [COMMAND_EXECUTION]: The skill uses npm rebuild to compile native Node.js addons, which involves the execution of arbitrary build scripts and potential network access to resolve dependencies.
  • [COMMAND_EXECUTION]: The skill ingest data from untrusted or external sources (the plugin cache and settings) and uses it to drive system-level changes and command execution without explicit sanitization or boundary markers.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 10, 2026, 03:50 AM
Security Audit — agent-trust-hub — claude-code-plugin-troubleshoot