telegram-cli
Pass
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPERSISTENCECOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads untrusted Telegram data (messages, names). This surface is mitigated by active character sanitization in the MCP server, data labeling for the AI model, and a two-step 'prepare-and-execute' approval cycle for all state-changing operations.
- [PERSISTENCE]: The tunnel setup script installs a systemd user service to keep the MCP server running. This is an intended persistence mechanism for the skill's local server functionality.
- [COMMAND_EXECUTION]: The Python implementation uses subprocess calls for virtual environment re-execution, and the MCP server uses controlled execution to invoke the CLI script. Both uses are standard for this type of utility.
Audit Metadata