The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data from Telegram chats and updates, creating a surface for potential indirect prompt injection. This is mitigated by schema validation of all API responses.
Ingestion points: The listRecentChats.ts script scans Telegram updates for chat information, and getChat.ts retrieves chat metadata.
Boundary markers: No explicit delimiters are used in the scripts for handling external data.
Capability inventory: The tools can send, edit, and delete messages, and perform network requests to the Telegram API.
Sanitization: All API responses are validated using Zod schemas in lib/telegram.ts, which enforces expected data types and strips undeclared fields.
[DYNAMIC_EXECUTION]: The cli.js entry point uses spawnSync to execute the skill's own TypeScript source or compiled JavaScript. This mechanism is restricted to the skill's own files and serves to support different JavaScript runtimes.
[UNVERIFIABLE_DEPENDENCIES_AND_REMOTE_CODE_EXECUTION]: The skill performs network requests to api.telegram.org to communicate with the Telegram Bot API. This is the primary purpose of the skill and targets a well-known service handled securely.

telegram