skills/karlorz/llm-wiki/wiki-remove/Gen Agent Trust Hub

wiki-remove

Pass

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to use various command-line utilities, including git, gh (GitHub CLI), rclone, and a specific skillwiki tool, to manage file deletions and synchronize state across local and remote environments.
  • [DATA_EXFILTRATION]: The instructions involve pushing data (deletion tombstones and metadata) to a remote Git repository (git push origin HEAD:main). This is a core feature of the skill designed to ensure file removals are durable across a distributed fleet.
  • [CREDENTIALS_UNSAFE]: Before performing network operations, the skill verifies environment permissions by checking gh auth status and git ls-remote. While it accesses sensitive repository URL information via git remote get-url, it does so for validation purposes within the user's established environment.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests data from a SCHEMA/index file to resolve vault paths. This represents an indirect ingestion surface where malicious content in the index could theoretically attempt to influence agent behavior, though the scope is limited by the deletion context.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 27, 2026, 11:30 AM
Security Audit — agent-trust-hub — wiki-remove