Load curated research from the two-tier research library. Tier 1: doctrine papers and dynamic deep dives in Supabase under research/papers/ and research/deep-dives/. Tier 2: quick-reference summary in docs/research/LIBRARY.md (load first for fast context). Use when the portfolio manager, asset analyst, or orchestrator needs evidence-based allocation rules or prior deep-dive research. Triggers: "research library", "paper notes", "tactical allocation doctrine", "cite research", "what do we know about", "prior research on".
Two-tier knowledge base. Load only what the current task needs.
docs/research/LIBRARY.md — 575-line condensed index of all doctrine papers with
Quick Reference tables, per-asset decision rules, and agent loading instructions.
Always load this first. It is the fastest path to citable allocation rules.
All papers and deep dives live in the documents table under research/ keys.
document_key patterns:
research/papers/{SLUG} — static doctrine papers (7 papers uploaded 2026-04-05)research/deep-dives/{TICKER}-{DATE} — single-asset or theme deep divesresearch/concepts/{SLUG} — timeless frameworks and conceptsresearch/themes/{SLUG}-{DATE} — market theme analysesLoad index:
python3 scripts/fetch_research_library.py
Fetch a specific note:
python3 scripts/fetch_research_library.py --key research/deep-dives/NVDA-2026-04-14
Filter by ticker or type:
python3 scripts/fetch_research_library.py --ticker NVDA
python3 scripts/fetch_research_library.py --type concept
Cache all to local scratch (optional session preload):
python3 scripts/fetch_research_library.py --cache data/agent-cache/research/
Do not paste entire papers or notes. Summarize decision rules only.