add-telegram
Warn
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: MEDIUMEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill adds a remote Git repository (https://github.com/qwibitai/nanoclaw-telegram.git) to fetch the necessary code for Telegram integration.\n- [REMOTE_CODE_EXECUTION]: The skill fetches and merges code from the telegram/main remote branch, which is then built and executed within the agent's environment.\n- [COMMAND_EXECUTION]: Several shell commands are used for system configuration and service management, including:\n
- Git commands for fetching and merging code.\n
- NPM commands (npm install, npm run build, npx vitest, npx tsx) for dependency management and script execution.\n
- Service management commands (launchctl kickstart, systemctl --user restart) to restart the agent service.\n- [INDIRECT_PROMPT_INJECTION]: The skill sets up an interface for Telegram messages, creating a surface for indirect prompt injection. The risk is explicitly mentioned when instructing users to disable Group Privacy, which allows the bot to see all messages in a group chat.\n
- Ingestion points: Incoming messages from the Telegram API processed by src/channels/telegram.ts.\n
- Boundary markers: None explicitly defined in the setup instructions to separate user messages from system instructions.\n
- Capability inventory: The merged code includes a TelegramChannel class that registers itself to the agent's communication layer.\n
- Sanitization: No specific sanitization or filtering of Telegram message content is described in the setup process.
Audit Metadata