What problem does it solve?
It eliminates the guesswork of choosing the correct agent or skill for a task by interpreting user intent, detecting domain signals, and selecting the best match (or requesting confirmation when ambiguous).
Core Features & Use Cases
- Intent-aware routing (informational vs operational): Avoids wasting tokens by answering questions directly when they are informational, and only routes when the request is operational or unclear.
- Multi-signal domain detection: Uses keyword scoring plus workspace inspection heuristics (e.g., Terraform, Docker, CI configs, dependency files) to infer the right domain.
- Pattern memory + named pipelines: Consults learned routing patterns and runs predefined “Named Pipelines” (e.g., full-audit, security-review) when trigger phrases match.
- Transparent decisioning & logging: Always announces a routing decision with confidence and logs the selection for learning and feedback.
Quick Start
Ask the AI to route this task, for example: "route this security and dependency audit for a repo with a GitHub Actions workflow and a package.json."
