The Agent Skills Directory

[SAFE]: The core logic of the skill is contained in the scripts/ directory. These Python scripts (debt_scanner.py, debt_prioritizer.py, debt_dashboard.py) use standard libraries to perform static analysis and data processing. No malicious operations such as hidden network requests, persistence mechanisms, or unauthorized file access were detected.
[CREDENTIALS_UNSAFE]: The assets/sample_codebase/ directory contains files with hardcoded placeholder credentials (e.g., sk_test_EXAMPLE_NOT_REAL in payment_processor.py and API_KEY = "abc123def456" in frontend.js). These are explicitly documented as intentional anti-patterns used to demonstrate the detection capabilities of the debt scanner and do not represent a security risk to the user or the agent's environment.
[PROMPT_INJECTION]: The skill ingests codebase data which may contain untrusted content. This creates a surface for indirect prompt injection.
Ingestion points: scripts/debt_scanner.py reads content from files within a target directory.
Boundary markers: None present in the reporting logic.
Capability inventory: Static code analysis (AST and regex) and report generation.
Sanitization: The scanner does not sanitize comments or string literals before including them in its output reports.
Note: This is a low-risk architectural characteristic common to code scanning tools.

tech-debt-tracker