skf-drop-skill

Pass

Audited by Gen Agent Trust Hub on May 11, 2026

Risk Level: SAFE
Full Analysis
  • [COMMAND_EXECUTION]: The skill performs recursive directory deletions when in 'purge' mode. This is the primary purpose of the skill. To mitigate risk, the skill implements a 'defense in depth' check to ensure deletions are restricted to the {skills_output_folder} and {forge_data_folder} directories. It also requires explicit user confirmation before any deletion occurs, unless 'headless mode' is specifically enabled by the user.
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by reading context-snippet.md files from skill directories and interpolating them into platform context files (e.g., AGENTS.md). While this is a standard orchestration behavior for the SKF platform, it represents a point where untrusted data enters the agent's context. The skill mitigates this by using boundary markers (<!-- SKF:BEGIN -->) and excluding deprecated versions from the rebuild process.
Audit Metadata
Risk Level
SAFE
Analyzed
May 11, 2026, 09:53 AM