link-to-im
Pass
Audited by Gen Agent Trust Hub on Apr 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: Fetches installation scripts and update packages from the developer's public GitHub repository (
d-wwei/Agents-To-IM). - [COMMAND_EXECUTION]: Spawns local processes for AI agents (
claude,codex,gemini) and executes development utilities such asnpm,esbuild, andffmpegas part of its core bridging functionality. - [REMOTE_CODE_EXECUTION]: Integrates self-update and self-healing modules (
update-kitandudd-kit) that can download and apply code changes from the upstream source upon user request or during installation. - [DATA_EXFILTRATION]: Transmits message data and session state to messaging platform APIs (Telegram, Discord, Feishu, QQ, WeChat) and optional third-party service providers (ElevenLabs, OpenAI) for voice processing.
- [PROMPT_INJECTION]: Processes untrusted external input from messaging platforms which creates an indirect prompt injection surface; the skill mitigates this by implementing a permission broker that requires manual user confirmation in the messaging app before the AI agent can execute sensitive tools.
Audit Metadata