Skip to main content

​
Summary

This starter turns the deep-research case study into a small project skeleton centered on planning, evidence collection, and citation-aware synthesis.

​
Status

starter

​
Why It Exists

Deep research is a flagship agent product shape in this repo. A small starter makes it easier to contribute traces, artifacts, and evaluation ideas later without copying a full external implementation.

​
Folder Structure

deep-research-agent-starter/
├── index.mdx
├── SOURCE_NOTES.mdx
└── src/
    ├── citation_formatter.py
    ├── research_loop.py
    └── research_review.py

​
Quick Start

This is a starter, not a finished product. The code sketch focuses on the core loop and leaves transport, UI, and persistence out of scope. For a repo-level smoke check, run python3 scripts/verify_example_projects.py from the repository root.

​
Included Sample Files

  • src/research_loop.py: the minimal planning, evidence, and draft-report loop
  • src/citation_formatter.py: a tiny reference surface for turning evidence rows into readable citations
  • src/research_review.py: a small post-research quality check for evidence coverage before publication

​
Constraints

  • No search adapter is implemented.
  • Citation formatting is illustrative.
  • Artifact persistence is not wired yet.

​
Next Steps

  • Add a real evidence store.
  • Add a report artifact writer.
  • Add evaluation cases for missing or weak evidence.