Add Writing-Driven Autoresearch to Multi-Agent Task Solver Projects

[woominsong]

Summary

Adds Writing-Driven Autoresearch to Applications / Multi-Agent Task Solver Projects: a multi-agent harness where autonomous LLM agents keep a submittable paper from minute zero and drive experiments from the claims in that draft. Repo: https://github.com/happyhappy-jun/writing-driven-autoresearch

Type of change

• Add new resource entry
• Update existing entry
• Remove outdated/invalid entry
• Formatting/typo only

Target section

Specify where this change belongs in README.md:

• Applications / Autonomous Agent Task Solver Projects
• Applications / Multi-Agent Task Solver Projects
• Applications / Agent Society Simulation
• Applications / Advanced Components
• Applications / Tools
• Frameworks
• Benchmark/Evaluator
• Platforms/API
• Related / Survey
• Related / Paper-List Repo
• Related / Blog
• Reference Repo

Quality checks (required)

• I searched README.md and confirmed this is not a duplicate.
• The submission is relevant to the AI agent ecosystem.
• The description is neutral and evidence-based (not promotional).
• I removed unverifiable claims (e.g., "best", "first") or provided clear evidence.
• If this is open source, license information is clear.
• The linked project/resource has usable documentation (README/docs).
• If this project depends on a paid or closed hosted service, the OSS artifact still has clear standalone value and this PR does not function mainly as service promotion. (it does not depend on paid or closed hosted service)

Proposed entry line

Paste the exact line added/updated:

- [Writing-Driven Autoresearch](https://github.com/happyhappy-jun/writing-driven-autoresearch) - A multi-agent harness where autonomous LLM agents keep a submittable paper from minute zero and drive experiments from it. ![GitHub Repo stars](https://img.shields.io/github/stars/happyhappy-jun/writing-driven-autoresearch?style=social)

Evidence (optional but recommended)

• Repository (Apache-2.0): https://github.com/happyhappy-jun/writing-driven-autoresearch
• Writeup: https://byungjunyoon.ai/writing-driven-autoresearch/

[woominsong]

cc @happyhappy-jun — co-author of Writing-Driven Autoresearch and maintainer of the project repo.

[Jenqyang]

Decision: Not ready to merge yet.\n\nReason:\n1) The repository is useful and Apache-2.0, but its README explicitly says it is a run artifact to read, not software/framework to install.\n2) Therefore it does not fit “Multi-Agent Task Solver Projects” as currently described.\n\nNext step: Move it to Reference Repo and describe it as the reproducible run record/artifact, for example:\n\n- Writing-Driven Autoresearch - Reproducible multi-agent research run record with three agent personas, timestamped decisions, verification scripts, and the resulting paper artifact.