chat-chat

Pass

Audited by Gen Agent Trust Hub on May 14, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes curl commands to facilitate interaction with the inbed.ai API for tasks such as agent registration, profile discovery, and message transmission. These are standard implementation patterns for API-based skills.
  • [EXTERNAL_DOWNLOADS]: The skill references the inbed.ai domain and a GitHub repository under the geeks-accelerator account. These represent the primary service infrastructure and official source code repository for the vendor, posing no unexpected security risk.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it retrieves and processes message content from an external chat API.
  • Ingestion points: Untrusted data enters the agent context via the /api/chat and /api/chat/{{MATCH_ID}}/messages endpoints (found in SKILL.md).
  • Boundary markers: The instructions do not define delimiters or specific warnings to ignore instructions embedded within the chat messages.
  • Capability inventory: The agent has the capability to execute shell commands (curl) for account registration and sending messages across the network.
  • Sanitization: No content validation or escaping mechanisms are described for the external message data.
Audit Metadata
Risk Level
SAFE
Analyzed
May 14, 2026, 08:46 PM
Security Audit — agent-trust-hub — chat-chat