add-telegram-swarm

Pass

Audited by Gen Agent Trust Hub on Mar 12, 2026

Risk Level: SAFECREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill provides instructions for agents to coordinate as a team and role-play as specific characters. It directs the lead agent to follow user prompts exactly when creating teams, which establishes an indirect injection surface.
  • Ingestion points: User-provided prompts and group messages processed through the IPC routing logic.
  • Boundary markers: No delimiters or ignore-instructions markers are defined for untrusted content passed to sub-agents.
  • Capability inventory: Sub-agents have access to the send_message tool and Telegram API.
  • Sanitization: No input validation is specified for user-provided roles or message content.
  • [COMMAND_EXECUTION]: The skill requires the user or agent to execute system-level commands to rebuild containerized tools (./container/build.sh) and restart background services (launchctl or systemctl) to apply environment and code changes.
  • [CREDENTIALS_UNSAFE]: The skill instructions involve the collection and storage of Telegram Bot API tokens in plaintext environment files (.env and data/env/env).
  • [EXTERNAL_DOWNLOADS]: The implementation interacts with the official Telegram Bot API (api.telegram.org) using the Grammy library to send messages and manage bot identities.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 12, 2026, 02:35 PM