troubleshooting-regression

Pass

Audited by Gen Agent Trust Hub on Mar 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes the claude CLI and suggests using pnpm for trace queries to facilitate debugging.\n
  • Evidence: Found in scripts/main.cjs where spawnSync is used to run the claude command.\n
  • Mitigation: The script calls spawnSync with shell: false, which mitigates shell-based command injection risks.\n- [DATA_EXFILTRATION]: The skill reads local debug logs from the user's home directory to identify error patterns.\n
  • Evidence: scripts/main.cjs accesses files in ~/.claude/debug/.\n
  • Mitigation: A filter function isIgnorableLine is implemented to skip lines containing authentication and credential-related keywords, preventing sensitive information from being processed or logged.\n- [PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection due to its handling of untrusted log data and user-provided prompts.\n
  • Ingestion points: Debug logs from ~/.claude/debug/ and the prompt argument in scripts/main.cjs.\n
  • Boundary markers: None present to distinguish data from instructions within the log files.\n
  • Capability inventory: Executes claude -p ... via spawnSync in scripts/main.cjs.\n
  • Sanitization: No sanitization or escaping of the prompt or log content before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 3, 2026, 02:59 AM