telegram

Pass

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill facilitates the creation of Telegram bots that ingest untrusted user text. The provided 'Automacao com IA' example in SKILL.md demonstrates passing this user text directly to an LLM without clear boundary markers or sanitization, creating a surface for indirect prompt injection attacks.
  • Ingestion points: User input is captured via update.message.text in assets/boilerplate/python/bot.py and ctx.message.text in assets/boilerplate/nodejs/src/handlers.ts.
  • Boundary markers: The implementation lacks explicit delimiters or 'ignore embedded instructions' prompts when interpolating user messages into the AI context.
  • Capability inventory: The skill's scripts and boilerplates include capabilities to send messages, perform network operations to the Telegram API, and manage chat memberships.
  • Sanitization: While basic html.escape is used for Telegram message formatting, no sanitization or validation is applied to user input before it is processed by the AI integration module.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 10, 2026, 08:02 AM
Security Audit — agent-trust-hub — telegram