Add batch mode for running research on multiple tickers

[juancadile]

Summary

• New bun batch <tickers.txt> command for headless batch processing
• Supports custom query templates with {TICKER} placeholder
• Each run creates timestamped output folder (outputs/YYYY-MM-DD_HH-MM-SS/)
• Generates individual JSON files per ticker + combined _combined.json
• JSON output includes: query, answer, tasks with tool calls, metadata

Usage

bun batch tickers.txt
bun batch tickers.txt --template "Analyze {TICKER} growth"
bun batch tickers.txt --output ./research

Output Structure

outputs/2025-01-12_14-30-45/
├── AAPL.json
├── GOOGL.json
├── NVDA.json
└── _combined.json   # All results in one file for LLM consumption

Test Plan

• [x] Tested with 41 tickers (~58 min total run)
• [x] Verified JSON output structure
• [x] Verified _combined.json generation
• [x] Typecheck passes

🤖 Generated with Claude Code

[juancadile]

Thank you for this repo, @virattt - it's super useful. I wanted to run a batch of an entire portfolio so I built this little feature. Hope all is good!