localhost(san-francisco) is next week π₯
Oct 14: the people who maintain, break and ship open source, in one room. No panel, no pitch.
Hosted by @sentry, co-organized with @superset_sh, @PlakarKorp, @OpenSpec_ & @commitvc.
Word of mouth almost filled the room β luma.com/hddco7hd
Founders & builders from @WorkOS, @nginx, @PingCAP, @mozilla, @strapijs, @zitadel, @StirlingPDF, Caution, @EntireHQ, Guepard, Confident AI, @papercompute, @modaicdev and more are coming π
π‘ ships its own compression model, Kompress-v2-base, on Hugging Face.
That's a wrap on OSSCAR Q2 Scaling π¬ Full ranking β osscar.dev
π‘ already used by tldraw & TanStack.
That's a wrap on OSSCAR Q2 Emerging π¬ Full ranking β osscar.dev
π‘ Launched days after Anthropic's closed-source Claude Design β the open, self-hostable, bring-your-own-model answer.
#1 tomorrow β osscar.dev
π₯ #2 OSSCAR Emerging Β· Q2 2026: omnigent
An open-source meta-harness giving Claude Code, Codex & Pi a common Python layer: policies, OS-level sandboxing, shareable live sessions.
π
Jun 2026 Β· π org this Q: ~6K stars (14x+) Β· 102 contribs (100x+) Β· ~53K dls (38x+)
π‘ Ships "Polly" (a multi-agent orchestrator that writes no code) & "Debby" (a Claude+GPT brainstorm partner).
#1 tomorrow β osscar.dev
π‘ Documents 71.5Γ fewer tokens per query vs reading raw files, tagging every edge EXTRACTED / INFERRED / AMBIGUOUS.
#2 tomorrow β osscar.dev
π₯ #3 OSSCAR Emerging Β· Q2 2026: Mnemosyne OSS @mnemosyne_oss
A zero-dependency, SQLite-backed memory system giving agents long-term context in one file.
π
Apr 2026 Β· π org this Q: ~1.4K stars (β) Β· 49 contribs (49x+) Β· ~40K dls (95x+)
π‘ Reports top scores on the LongMemEval & BEAM long-term-memory benchmarks.
#2 tomorrow β osscar.dev
6οΈβ£ TinyHumans (@tinyhumansai) β OpenHuman, a local-first personal intelligence that remembers your life
5οΈβ£ DocLang β an AI-native, ISO-standard markup mapping documents, audio and video onto LLM tokens








