send-message

Pass

Audited by Gen Agent Trust Hub on Nov 7, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONPERSISTENCE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill enables the agent to read from and write to a shared Telegram supergroup, which introduces a surface for indirect prompt injection from external users.
  • Ingestion points: The agent fetches message content, reactions, and citations from Telegram via client.get_messages as part of the 'Supergroup-First Methodology'.
  • Boundary markers: There are no specified delimiters or instructions to ignore potential commands within the ingested Telegram content.
  • Capability inventory: The skill allows shell execution via bun and has file system access for managing sessions.
  • Sanitization: No sanitization or validation logic is provided for content retrieved from the Telegram API.
  • [PERSISTENCE]: The skill includes a 'Self-Evolving Skill' block that explicitly directs the agent to 'fix this file immediately' if issues or workarounds are identified. This self-modification capability allows for the persistence of behavioral changes across sessions, which could be exploited to permanently alter the agent's logic if it is influenced by malicious input during a task.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 7, 2026, 02:12 AM