The Agent Skills Directory

[SAFE]: The skill provides extensive documentation on Telegram Bot development with a strong focus on security best practices.
[CREDENTIALS_UNSAFE]: The skill explicitly prohibits hardcoding tokens and secrets, recommending the use of environment variables instead (see SKILL.md and references/webhooks.md).
[DATA_EXFILTRATION]: No patterns of unauthorized data access or exfiltration were found. The code snippets for handling user data include mandatory verification steps.
[PROMPT_INJECTION]: There are no instructions that attempt to bypass AI safety guidelines or override agent behavior. Terms like 'CRITICAL' and 'IMPORTANT' are used correctly within the context of developer guidance.
[REMOTE_CODE_EXECUTION]: No remote code execution patterns, such as piped shell commands or dynamic execution of untrusted scripts, were detected.
[INDIRECT_PROMPT_INJECTION]: While the skill describes handling untrusted data from Telegram, it provides specific mitigation strategies including:
Ingestion points: aiogram handlers and webhook endpoints receive data from the Telegram Bot API.
Boundary markers: Instructions explicitly require validating the X-Telegram-Bot-Api-Secret-Token and verifying cryptographic hashes for authentication data.
Capability inventory: The skill's patterns are limited to standard bot operations (sending messages, processing payments via official providers) and do not involve dangerous system capabilities like eval or file system writes.
Sanitization: Implementation guides for verify_telegram_auth and validate_mini_app_init_data ensure data integrity using hmac.compare_digest for constant-time comparison.

telegram