Skills
Navigate
skills/iblai/vibe/iblai-vibe-local-llm/Gen Agent Trust Hub

iblai-vibe-local-llm

Pass

Audited by Gen Agent Trust Hub on Sep 10, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches brand guidelines and configuration documentation from the official iblai GitHub repository.\n- [COMMAND_EXECUTION]: Instructs the developer to implement commands for system resource monitoring (using nvidia-smi) and local service management (Ollama daemons).\n- [REMOTE_CODE_EXECUTION]: Documents the standard installation procedure for Ollama using a shell script from a well-known service domain.\n- [INDIRECT_PROMPT_INJECTION]: The skill defines an interface for processing untrusted chat history through local model providers.\n
  • Ingestion points: The ollama_chat and ollama_chat_stream command signatures in references/tauri-commands.md receive message arrays.\n
  • Boundary markers: No specific delimiters are mandated in the contract within SKILL.md or references/architecture.md.\n
  • Capability inventory: The proposed architecture in references/architecture.md includes capabilities for local network communication and subprocess execution for tool installations.\n
  • Sanitization: The documentation does not specify input sanitization requirements for the inference stream.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 10, 2026, 11:24 AM