Google Research open-sources PaperOrchestra, a multi-agent framework that writes AI papers end-to-end
adnan_hashmi · x · 2026-11-16
Google Cloud AI Research released PaperOrchestra, an open-source (Apache-2.0) multi-agent framework that turns unconstrained pre-writing materials — idea summaries, experimental logs, LaTeX templates, conference guidelines — into submission-ready papers.
The pipeline runs in five steps:
- Outline Agent: produces a JSON outline, visualization plan and research graph
- Plotting Agent: generates figures (backbone: PaperBanana)
- Literature Review Agent: combines web search with the Semantic Scholar API to identify and verify references, drafting intro/related work plus a .bib citation bank
- Section Writing Agent: drafts the full manuscript
- Iterative Refinement: an AgentReview evaluator and content-refinement agent loop improvements until the final PDF passes checks
The repo hit 920 stars shortly after release.
More from coding & agent
- Stanford CS329Z draws 100+ students in debut, publishes full syllabus with industry API credits — FinanceYF5 · 2026-11-16
- SWE-bench co-creator designs Stanford CS329Z HW2 on Terminal-Bench-style agent evals — FinanceYF5 · 2026-11-16
- Stanford CS329Z open-sources HW1: build an agent harness from scratch for under $5 — FinanceYF5 · 2026-11-16
- Stanford opens CS329Z agent engineering course: build a Claude Code-like agent for under $5 — FinanceYF5 · 2026-11-16
- Open-source Ontology Atlas gives humans and coding agents one shared Markdown codebase ontology — adnan_hashmi · 2026-11-16
- GH-AWTrace: 25,014 GitHub Agentic Workflows Linked to 4.1M Run Records — ETSMTL · 2026-11-16