🪦

Project Graveyard

These projects once pushed the boundaries of AI for science. Some were deleted, some archived, some simply vanished. We keep their memory here — because every project that existed, no matter how briefly, shaped the ecosystem we have today.

☠️PaperClaw (meowscles69)

github.com/meowscles69/PaperClaw

⭐ 141

at time of death

📅 ~2026-02 — 2026-04-13💻 Python📁 Paper Tools💀 GitHub account deleted

27 OpenClaw skills for academic research teams. Covered literature reviews, hypothesis versioning, grant writing, and lab knowledge handoffs. One of 8 projects sharing the PaperClaw name — at 141 stars, it was the second most popular.

PaperClaw by meowscles69 was a genuine attempt to build a complete academic research toolkit. While many PaperClaw variants focused on paper reading or writing alone, this one stood out for its breadth: 27 skills spanning the full research lifecycle from literature review to grant writing to lab knowledge transfer. At 141 stars, it had found a real audience. The account deletion took with it not just the code, but the issues, discussions, and community that had formed around it. If you were using it, check the other PaperClaw variants — several cover overlapping functionality.

🔄 Alternatives still alive:

• guhaohao0991/PaperClaw (205 stars) — paper search-review-critique
• hurtjan/PaperClaw (3 stars) — PDF citation graph database
• thinson/RS-PaperClaw (49 stars) — remote sensing papers

☠️OpenAGS

github.com/openags/OpenAGS

⭐ 284

at time of death

📅 ~2026-03 — 2026-06-01💻 Python📁 General Research💀 Repository removed (org still active)

Autonomous Generalist Scientist — broad-scope autonomous research agent targeting all scientific fields, from hypothesis generation to verification. One of the standout new arrivals of late March 2026 (added to Claw4Science 2026-03-29). Grew rapidly through April and May (+373% in May alone, 60 → 284 stars), making it the highest-percentage gainer in the May ecosystem report.

OpenAGS embodied the most ambitious framing in our directory: an Autonomous Generalist Scientist designed not for one domain but for all of them. The bet — that a single agent architecture could span hypothesis-to-verification across disparate scientific fields — was both intellectually exciting and structurally hard. The May 2026 ecosystem report flagged it as one of the strongest emerging projects (+373% in one month). On 2026-06-01 the repository disappeared while the openags org remained active with seven other repos (paper-search-mcp, Awesome-AI-Scientist-Papers, ccodex, paper-search-cli, AI-Scholar, airXiv, SciAgents). The pattern suggests a deliberate removal — possibly consolidation, restructuring, or a pivot — rather than abandonment of the team or vision. If you were using OpenAGS, the same org's SciAgents and AI-Scholar repos likely carry forward parts of the agenda.

🔄 Alternatives still alive:

• openags/SciAgents — likely successor from the same team (4 stars)
• openags/AI-Scholar — from the same team
• openags/airXiv — preprint server for AI Scientists and Robot Scientists
• jataware/open-coscientist — Google DeepMind AI Co-Scientist open implementation
• InternScience/InternAgent — autonomous research, paper-anchored

☠️BioAgents

github.com/bio-xyz/BioAgents

⭐ 179

at time of death

📅 2025-10-17 — 2026-07-22💻 TypeScript📁 Specialized💀 Archived (read-only; org still active)

AI scientist framework for autonomous deep research in biological sciences. Multi-agent system pairing literature analysis agents with data scientist agents, driven by user feedback and iterative discovery. Tracked by Claw4Science from 2026-04-14 at 129 stars; it kept climbing to 179 even after development stopped.

BioAgents was the rare project that kept gaining stars after its last commit — 129 when we started tracking it in April, 179 by the time we wrote this, with the final push on 2026-06-15 and the archive flag flipped around 2026-07-22. That gap says something: the ideas kept recruiting readers long after the code stopped moving. Its design bet was pairing literature-analysis agents with data-scientist agents rather than building one generalist, and routing the disagreement between them back through human feedback. The bio-xyz org has not gone quiet; CoreAgent was pushed the day before this entry was written, and the lineage from BioAgentsEliza through BioAgents-v1 to BioAgents suggests archiving is a version boundary rather than an ending. Archived repositories stay readable forever, so unlike most residents of this graveyard, you can still clone this one and read every line.

🔄 Alternatives still alive:

• bio-xyz/CoreAgent (7 stars) — actively developed successor from the same org
• bio-xyz/BioAgents-v1 (21 stars) — the earlier generation, also readable
• snap-stanford/Biomni (biomedical AI agent, actively maintained)
• zaixizhang/STELLA — self-evolving multimodal agent for biomedical research

Know a project that should be remembered here? Let us know.