Fetching from the wire…
01
02
03
04
05
06
07
08
Public story · 2026-08-05 · high
It added 3,601 stars to reach 21,018 and shipped tagged macOS alpha builds on August 3 and 4, per its GitHub release history.
Why now: Openwork's release history places v0.18.13 and v0.18.14 on August 3 and 4, right before its star count crossed 21,000 on August 5.
Openwork has climbed to 21,018 GitHub stars and 2,060 forks after adding 3,601 stars this stretch, per its repository page.
The project describes itself flatly as "the open-source alternative to Claude Cowork (powered by opencode)." The description credits opencode as the engine underneath the interface, not a rebuild from scratch. That's the stake for anyone picking an agent runner: a same-week open clone is shipping faster than the original's public changelog.
Release cadence backs the growth up. Openwork shipped v0.18.13 on August 3 and v0.18.14 the next day. A tagged macOS alpha build, 0.18.14-alpha.2119, landed on both of those dates, according to its GitHub release history.
Two related projects show the same pressure from other angles. Anthropic patched a permission-escalation hole in the Claude Agent SDK where a crafted skill name could grant itself extra tool access. Uber separately open-sourced an agent security benchmark spanning 133 MCP servers and 17 attack techniques.
Each link below shares sources, entities, or timing with this story.
Every write, send or shell command needs a human okay, and unattended runs queue their asks in an inbox instead of acting.
different-ai/openwork added 3,601 stars this week (2,060 forks), describing itself flatly as "the open-source alternative to Claude Cowork (powered by opencode)." v0.18.13 on August 3, v0.18.14 on August 4, plus tagged macOS alphas the same days. A proprietary Anthropic surfac...
different-ai/openwork is a macOS/Windows/Linux desktop app built on opencode letting non-technical staff work with agents on local files, shipping v0.18.12 on July 30. YC's launch post cites over 150K downloads, with on-prem deployment and support for 50+ LLM providers as the...
The open-source project replaces vector search with ls and find commands, and its LoCoMo scores go from 24-57% to 80-83% accuracy.
Serverless compute, bug-fixing agents, retrieval models and moderation APIs each got a free replacement between August 4 and 6.
The open-source tool now sits at 98.9k GitHub stars and pairs a skill with a proxy that compresses code, logs and JSON before they hit the model.
MindPattern daily
One email a day at 7 AM. Sources and a take on every story. Unsubscribe anytime.