error-debugging-error-trace
Pass
Audited by Gen Agent Trust Hub on Oct 26, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted codebase data and logs for static analysis, creating a surface for indirect prompt injection if the analyzed files contain malicious instructions targeted at the agent.\n
- Ingestion points: The ErrorTrackingAnalyzer in resources/implementation-playbook.md reads all project files matching common source extensions using Path.rglob.\n
- Boundary markers: The instructions lack explicit delimitation or 'ignore instructions' warnings for the content being processed by the analyzer script.\n
- Capability inventory: The skill provides functionality for network transmission (fetch, Slack webhooks), recursive file system reading, and structured logging.\n
- Sanitization: resources/implementation-playbook.md includes a sanitizeEvent function to redact common sensitive keys (password, token, etc.), and SKILL.md provides high-level instructions to avoid logging personal data.\n- [COMMAND_EXECUTION]: The skill includes code for recursive file system traversal and content reading to perform its diagnostic functions.\n
- Evidence: resources/implementation-playbook.md contains a Python class ErrorTrackingAnalyzer that uses Path(project_path).rglob('*') to traverse the project directory and file_path.read_text() to inspect file contents.
Audit Metadata