The Agent Skills Directory

[SAFE]: The skill provides automated link verification and repair functionality. Its operation involves scanning project files and testing URLs, which aligns with its stated purpose.\n- [SAFE]: The included Python script scripts/profile_store.py manages persistent user settings. It incorporates security measures to prevent the storage of sensitive information like tokens or passwords by using keyword filtering and requiring explicit user confirmation for writes.\n- [SAFE]: Structural validation of the skill is performed by scripts/validate_skill.py, which uses safe loading practices (yaml.safe_load) for configuration files and checks for hygiene issues like private user paths.\n- [INDIRECT_PROMPT_INJECTION]: The skill ingests content from external sources (project files like HTML and Markdown) to extract URLs. While this creates a theoretical attack surface for indirect prompt injection, it is the primary function of the skill and is managed by its specific task scope, clear boundaries against automated actions, and confirmation workflows. Evidence: Ingestion points: Project files (HTML/Markdown/JS); Boundary markers: Instructions to distinguish error codes and avoid guessing URLs; Capability inventory: File write access to fix links, network access for checking external URLs; Sanitization: Standard link parsing is used without explicit sanitization logic for data processing.

lov-fix-broken-links