How jameslederman/alpha-daemon works
AlphaDaemon is point-in-time market research infrastructure. A run is created with a symbol and an as-of cutoff, then a Temporal workflow gathers evidence, plans research, retrieves and ranks SEC evidence, produces findings, and synthesizes a recommendation. A separate fundamental-analysis workflow prepares a filing corpus, lets a bounded analyst query SEC filings, news, and price history, and persists its result. The README also describes forecast assumptions and deterministic valuation as longer-term goals; those stages are not represented as implemented components. Some subsystem paths and wiring are based on README descriptions because their source was not sampled.
Main entry points: activities.py Research Activities · tools.py Domain Tools · storage.py Corpus Storage · workflows.py Research Workflow
Components 20
Run Entry
- Run Launcher
src/alpha-daemon/run_workflow.py - Research API
src/alpha-daemon/api.py
Durable Orchestration
- Worker
src/alpha-daemon/worker.py - Research Workflow
src/alpha-daemon/workflows.py - Fundamental Workflow
src/alpha-daemon/workflows.py - Research Activities
src/alpha-daemon/activities.py
Research Capabilities
- Fundamental Analyst
src/alpha-daemon/fundamental_analyst.py - Domain Tools
src/alpha-daemon/tools.py - Market Data Provider
src/alpha-daemon/market_data.py - LLM Client
src/alpha-daemon/llm.py - Research Models
src/alpha-daemon/models.py
Evidence Infrastructure
- SEC Filing Access
src/alpha-daemon/edgar.py - Chunking and Retrieval
src/alpha-daemon/retrieval.py - Corpus Storage
src/alpha-daemon/storage.py
Other components
- Research User
- Temporal Service
- Massive API
- SEC EDGAR
- PostgreSQL Corpus
- Amazon Bedrock
Connections 25
- Research User → Run Launcher: starts run
- Run Launcher → Temporal Service: starts workflow
- Research API → Temporal Service: submits run
- Worker → Temporal Service: registers tasks
- Temporal Service → Research Workflow: dispatches
- Temporal Service → Fundamental Workflow: dispatches
- Research Workflow → Research Activities: executes activities
- Research Workflow → Research Activities: saves run
- Fundamental Workflow → Research Activities: prepares corpus
- Fundamental Workflow → Fundamental Analyst: invokes analysis
- Fundamental Analyst → Research Activities: calls tools
- Research Activities → Domain Tools: uses capabilities
- Research Activities → LLM Client: generates text
- LLM Client → Amazon Bedrock: invokes model
- Domain Tools → Market Data Provider: fetches market data
- Market Data Provider → Massive API: requests bars and news
- Domain Tools → SEC Filing Access: fetches filings
- SEC Filing Access → SEC EDGAR: requests SEC data
- SEC Filing Access → Corpus Storage: reads and caches filings
- Domain Tools → Chunking and Retrieval: chunks and ranks
- Research Activities → Chunking and Retrieval: retrieves evidence
- Domain Tools → Corpus Storage: reads and writes corpus
- Research Activities → Corpus Storage: loads evidence
- Corpus Storage → PostgreSQL Corpus: persists corpus and runs
- Chunking and Retrieval → Amazon Bedrock: embeds evidence