GCP-First Agentic AI Project
Stock Intelligence
A multi-agent market analysis platform with specialized technical, sentiment, macro-risk, and portfolio agents.
Architecture
A2A orchestration
MCP tool contracts
Cloud Run + Firebase
Sector Explorer
Sector Rotation
Decision Console
Idle
Submit a symbol to generate a multi-agent market intelligence report.
Agent Diagram
How The Agent Team Reaches A Decision
The workflow moves from deterministic signal generation into supervisor synthesis, then ends with risk-aware decision support and report generation.
User selections and notes seed the analysis context.
Each agent scores a different dimension of conviction.
Signals are translated into position stance and execution discipline.
Supervisor synthesis resolves conflicts and produces the final summary.
Decision Support KPIs
How To Read The Output
These metrics are designed for explainable decision support. They summarize how the agent team scores conviction, risk, and execution discipline.
Confidence Score
The supervisor agent combines technical, sentiment, macro, and portfolio signals into a final confidence value from 0 to 100.
Signal Scores
Each specialist agent produces its own score so you can see where conviction is strong and where the recommendation is being weakened.
Risk Guardrails
Position sizing, human review, and risk disclosure are always shown so the system stays grounded in controlled execution rather than blind prediction.
A2A Workflow Trace
The workflow section shows how agents hand off work. That makes the orchestration transparent and easier to explain in research reviews and demos.