gifgrep

Pass

Audited by Gen Agent Trust Hub on Mar 30, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill depends on a binary that must be installed from a third-party repository (github.com/steipete/gifgrep). It provides instructions for installation via Homebrew and the Go toolchain.
  • [COMMAND_EXECUTION]: The skill uses the gifgrep tool to perform local file system operations, including downloading media to the user's downloads folder and invoking system-level 'reveal' commands to open files in the Finder.
  • [DATA_EXFILTRATION]: The skill connects to external API providers (Tenor and Giphy) to fetch content. It facilitates the use of sensitive environment variables, such as GIPHY_API_KEY and TENOR_API_KEY, to authenticate these network requests.
  • [PROMPT_INJECTION]: The skill handles untrusted metadata (e.g., GIF titles and tags) from external providers, which serves as a potential surface for indirect prompt injection. Ingestion occurs during GIF provider search operations. No specific sanitization or boundary markers are described for this external content before it is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 30, 2026, 01:09 AM
Security Audit — agent-trust-hub — gifgrep