18ac47f5d6
Standardized input/output contract for all agents, enabling pluggable composition and benchmarking. - AgentInput/AgentOutput Pydantic schemas (5-tier rating, confidence, targets) - BaseAgent abstract class with analyze(input) -> output contract - AgentRegistry for pluggable agent discovery - Existing analysts refactored to implement BaseAgent - Agent benchmarking: compare outputs across different LLM backends Closes #264 |
||
|---|---|---|
| .. | ||
| analysts | ||
| managers | ||
| researchers | ||
| risk_mgmt | ||
| trader | ||
| utils | ||
| __init__.py | ||
| base_agent.py | ||
| benchmark.py | ||
| registry.py | ||