Interceptor

Pass

Audited by Gen Agent Trust Hub on Aug 15, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses tools like interceptor, curl, magick, ffmpeg, and bun for automation and data processing. These provide high-capability system access required for Computer Use tasks.
  • [EXTERNAL_DOWNLOADS]: The Update workflow fetches source code from github.com/Hacker-Valley-Media/Interceptor and installs dependencies via bun install. This is for maintaining the tool from its official repository.
  • [REMOTE_CODE_EXECUTION]: Instructions describe building and installing binaries, re-signing code, and registering LaunchAgents. These are necessary for the tool's infrastructure setup.
  • [PROMPT_INJECTION]: The skill uses instructions to enforce environment isolation. It also possesses an indirect prompt injection surface when reading external webpages. Ingestion points: interceptor read and interceptor text verbs. Boundary markers: Use of semantic refs and markdown rendering is encouraged. Capability inventory: High-privilege shell access, filesystem operations, and network control. Sanitization: Relies on scoped reads rather than active content filtering.
  • [DATA_EXFILTRATION]: Uses curl to send local notifications to http://localhost:31337/notify upon invocation. This is for local user feedback within the execution environment.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 15, 2026, 06:48 PM
Security Audit — agent-trust-hub — Interceptor