At enterprise scale, your AI agent has access to hundreds of plugins with overlapping descriptions. The LLM reads every plugin description in the context window and picks the most probable match — but “most probable” and “correct” aren’t the same thing. Misrouted tool calls compound: wrong plugin, wrong parameters, wrong outcome.
When an AI agent needs to choose a tool to call, multiple similarly named tools can confuse the AI agent.
The Manifest Generator inspects the code in your plugins to optimize them for tool selection. You can learn more about how this is done in our Conversation Triggers docs .
When your users chat with your AI Assistant , the Manifest Generator takes the following steps:
