The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses curl and jq as its primary execution engines to interact with the Telegram API. These are used strictly for their intended purposes: making authenticated HTTP requests and parsing JSON responses.
[CREDENTIALS_UNSAFE]: The script manages bot tokens by reading them from a local configuration file (~/.config/telegram/config) or environment variables. It implements a security best practice by passing tokens to curl through the --config - flag on stdin, ensuring tokens do not appear in process monitoring tools like ps or top.
[DATA_EXFILTRATION]: While the skill sends data to Telegram's servers, this is its primary stated purpose. The SKILL.md includes explicit warnings about data leaving the local machine and advises against sending sensitive workspace or customer data.
[INDIRECT_PROMPT_INJECTION]: The read and ask commands ingest external content from the Telegram API. The skill mitigates risks by requiring specific TELEGRAM_APPROVER_IDS for group chats and validating that incoming messages originate from allowed chat IDs.

telegram-bot-messaging