How huainguyn/adaptive-recommendation-agent works
This repository is a standalone Adaptive Recommendation Agent demo. Its main workflow accepts a Consistency Result JSON, validates it, applies rule-based per-issue decisions, and returns a Recommendation Result JSON with a suggested next action; it does not execute that action. It also exposes a FastAPI endpoint and a terminal sample runner. The README documents coordinator integration and recommendations to request further analysis. The graph reflects the shown API, workflow, and recommender code; it does not infer calls to other agents.
Main entry points: agent.py Workflow agent · interface.py FastAPI interface · schema.py Request and result contracts · main.py Sample runner
Components 8
Entry points
- FastAPI interface
interface.py - Sample runner
main.py
Recommendation workflow
- Workflow agent
agent.py - Request validation
agent.py - Rule-based decisions
recommender.py
Data contracts
- Request and result contracts
schema.py
Other components
- API caller
- Terminal user
Connections 10
- API caller → FastAPI interface: submits JSON
README.md - FastAPI interface → Workflow agent: invokes
interface.py - Workflow agent → Request validation: validates input
agent.py - Request validation → Request and result contracts: parses request
agent.py - Workflow agent → Rule-based decisions: runs recommendations
agent.py - Rule-based decisions → Request and result contracts: builds result
recommender.py - Workflow agent → FastAPI interface: returns result
interface.py - Terminal user → Sample runner: runs demo
README.md - Sample runner → Workflow agent: invokes
main.py - Workflow agent → Request and result contracts: returns response model
agent.py