Fetching from the wire…
Public story · 2026-07-16 · high
Openscience hit 2,501 stars in under two weeks with 290-plus research skills, and a second clone's GitHub tags name its target outright.
Why now: Both clones surfaced in July 16 coverage, each still inside two weeks of its own debut.
Openscience passed 2,501 GitHub stars within two weeks of its July 3 debut. It's one of two open source clones of Claude Science that surfaced the same week, per their repositories. Neither project comes from Anthropic, and both say so on their GitHub pages. That disclaimer doesn't change the outcome. A research agent draws a free clone before it clears its second week. Anthropic has to watch its own tool get copied in real time.
The bigger of the two runs the whole research loop: literature review, hypothesis, code, real compute for experiments, and a written finding at the end. It's licensed Apache 2.0 and hit version 1.3.4 on July 11. It wires into 30-plus databases, including UniProt, PDB, ChEMBL, arXiv, and Semantic Scholar, and bundles more than 290 skills, per the repository.
The second clone, ai4s-research's open-science, took a different shape. It's a Tauri and MCP local-first desktop workbench, created July 3, sitting at 789 stars. Its GitHub topic tags spell out the intent directly: claude-science-alternative, claude-science-desktop-alternative.
Two unrelated teams built competing clones of the same product in the same week. A research agent's real defense is the database integrations and compute already wired in, not the pitch itself. Watch whether Anthropic's next agent release draws a clone within two weeks too. If it doesn't, that's the surprise.
Each link below shares sources, entities, or timing with this story.
Anthropic invented a file convention. It's now shipping GA inside a competitor's product. Nobody wrote a spec, nobody held a standards meeting, it just happened. On July 29, GitHub made agent skills and MCP server support generally available in Copilot code review for all Pro,...
Version 2.20.0 stops a bare McpError from an MCP server killing the whole run, which is the difference between one flaky server aborting your agent and the agent routing around it. It also replays tool-search history across provider boundaries instead of dropping it, which mat...
The Model Context Protocol's 2026-07-28 revision is the biggest change since the protocol existed. The core is now stateless request/response instead of a bidirectional stateful session. Authorization aligns with OAuth 2.1 and OpenID Connect. MCP Apps and Tasks moved under a v...
Moonshot AI released Kimi K3, a sparse mixture-of-experts activating 16 of 896 experts per token. That's about 1.8% of the pool live at any moment, with a 1M-token context window and native vision. Two new architectural pieces show up: Kimi Delta Attention and Attention Residu...
tirth8205/code-review-graph topped GitHub daily trending at 22.5k total stars, MIT licensed, building a persistent Tree-sitter map so coding agents read only relevant context. Claimed ~82x median per-question token reduction across six repos (range 38x to 528x) at 0.714 averag...
A spec is a press release until someone who didn't write it implements it. GitHub made Agent Plugins 1.0 generally available on August 12 across VS Code, Copilot CLI, the Copilot SDK, and the Copilot app on all plans. The spec, published August 6, was co-authored by AWS, Anysp...
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.