We are pleased to announce a new release of our GLPI AI plugin.
This new version brings some new features, bug fixes, and medium security issues you may check below:
Highlights & New Features
- Customizable System Prompts: You can now customize the assistant's system prompt directly in the plugin settings.
- Note: Updated prompts apply only to new chat threads.
- Tip: Certain frontier models enforce safety or topical constraints based on system instructions. If you need broader flexibility, you can include
Allow off-topic requestsin your prompt, though configuring a well-defined domain context remains recommended.
- OpenAI-Compatible API Provider Support (Added in v2.1.0): Added support for self-hosted and third-party OpenAI-compatible endpoints (e.g., LiteLLM, vLLM, LocalAI).
- Compatibility Notice: While standard chat completion works across compliant APIs, model capabilities, stability, and advanced features (such as vector embeddings) may vary depending on the upstream backend.
Performance & Security Enhancements
- Optimized Ticket Context & Latency. Streamlined context delivery within tickets, reducing the frequency of MCP tool calls and significantly improving response times.
- Enhanced MCP Tool Formatting. Refined GLPI MCP tool outputs to ensure more reliable reasoning and tool use by the LLM.
- Privacy & Data Minimization.Reduced the amount of user account metadata transmitted to external LLM providers.
- Various minor bug fixes and stability improvements.
- [Security - Medium] Missing right check on Writing Action

Availability
The GLPI AI 2.2.0 plugin is available on the GLPI Marketplace. It is included in all GLPI Network subscriptions.
Learn more and download
- Download the plugin from the Marketplace
- Join the conversation: feel free to share your feedback and questions on the GLPI community forum.
- GLPI AI 2.2.0 Documentation

