The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The file src/tg_cli/config.py contains a hardcoded API hash (_DEFAULT_API_HASH). This value is the public constant for the official Telegram Desktop application (API ID 2040). While its inclusion facilitates ease of use, hardcoded credentials in source code are generally flagged as a risk. In this context, it is mitigated as it is associated with the skill's primary functionality for interacting with a well-known service.
[PROMPT_INJECTION]: The skill presents an indirect prompt injection surface (Category 8). (1) Ingestion points: External data from Telegram messages enters the agent's context through fetch_history in src/tg_cli/client.py. (2) Boundary markers: Findings show that messages are output to the agent context in raw or structured YAML formats without explicit delimiters or instructions to ignore embedded commands. (3) Capability inventory: The skill has significant capabilities including sending, editing, and deleting messages (tg send, tg edit, tg delete in src/tg_cli/cli/tg.py), which could be abused if an agent obeys a malicious instruction contained within a processed message. (4) Sanitization: Incoming message content is not sanitized or filtered before being displayed or stored.

tg-cli