skills/iblai/api/iblai-api-agent-mcp/Gen Agent Trust Hub

iblai-api-agent-mcp

Pass

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [SAFE]: The skill is designed to interact exclusively with official vendor infrastructure at https://api.iblai.app/dm. All network operations are directed to the vendor's own API for managing agents and MCP servers.
  • [EXTERNAL_DOWNLOADS]: The documentation references the official iblai/iblai-mcp GitHub repository for ready-made server configurations. This is a trusted vendor resource for this skill.
  • [PROMPT_INJECTION]: The skill manages tools that ingest data from external MCP servers, which is an inherent surface for indirect prompt injection. The skill documentation describes a clean_output sanitization feature (enabled by default) to mitigate risks associated with processing external content.
  • Ingestion points: Data enters the agent's context through external MCP server responses via SSE, WebSocket, or HTTP transport protocols described in SKILL.md.
  • Boundary markers: The clean_output flag in the server metadata is used to control the processing of output, though explicit instruction-level boundary markers are handled by the underlying platform.
  • Capability inventory: The skill allows the agent to register new servers, manage OAuth connections, and update its own tool wiring and settings via REST API calls.
  • Sanitization: The clean_output setting defaults to true, which automatically strips HTML from server responses to prevent markup-based injection.
  • [SAFE]: The skill implements strong credential security by requiring that credentials and custom headers are masked when retrieved via the API and by explicitly warning developers never to echo secrets in chat or commit them to files.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 27, 2026, 11:40 AM
Security Audit — agent-trust-hub — iblai-api-agent-mcp