sync-aiblueprint-with-claude

Pass

Audited by Gen Agent Trust Hub on Mar 3, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: Uses Bash tools including cp, diff, ls, and mkdir to perform file operations and directory management within the user's home directory.- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted Markdown and script files from a local repository and moves them into the agent's active configuration path without content validation.
  • Ingestion points: All files within $CWD/claude-code-config/.
  • Boundary markers: None; the skill copies the source files verbatim into the target directory.
  • Capability inventory: The skill possesses file write permissions, directory creation capabilities, and the ability to execute shell commands.
  • Sanitization: No sanitization or safety filtering is applied to the content of the files being synchronized.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 3, 2026, 07:43 AM