The Agent Skills Directory

[SAFE]: The core scanning script scripts/scan_workspace_cleanup.py uses standard libraries to perform directory traversal and metadata collection; it contains no logic for file deletion, moving, or content modification.
[SAFE]: No network operations or data exfiltration vectors were identified; the skill does not use libraries like requests or system tools like curl.
[SAFE]: The skill documentation and automation templates in SKILL.md and references/automation-prompts.md establish strict safety boundaries, explicitly prohibiting the AI from performing destructive actions.
[SAFE]: Indirect Prompt Injection analysis: (1) Ingestion points: scripts/scan_workspace_cleanup.py reads local file and directory names. (2) Boundary markers: The script returns results in a structured JSON or ranked text format. (3) Capability inventory: The skill has no capabilities for network access, file writing, or subprocess execution. (4) Sanitization: While filenames are not sanitized, the tool's read-only nature prevents accidental instruction execution from poisoned filesystem metadata.
[SAFE]: No obfuscation, hardcoded credentials, or persistence mechanisms were found across the skill's file set.

workspace-cleanup-audit