Skim
Free, open-source AI email client for Windows
Start new thread

Skim v1.0.6 — "Bring Your Own Brain" 🧠

by•

The one where Skim stops being picky about which AI you talk to, and the whole app goes on a diet at the same time. Efficiency and choice, in one download. You're welcome.

✨ New

  • Bring your own AI endpoint (OpenAI-compatible). 🔌 A third provider joins Anthropic and OpenRouter: point Skim at any OpenAI-compatible API — Ollama, LM Studio, vLLM, your homelab gateway, whatever. Just a Base URL and a Model ID; the API key is optional (leave it blank for local servers that don't want one). No key = no Authorization header, so your Ollama box stops complaining. (Closes .)

  • AI chat that can actually breathe. 🫁 The command palette (Ctrl+K) can now expand to near-fullscreen for reading long answers — one toggle in the corner, only in chat mode. Repeat Ctrl+K flips it open/closed instead of nuking the palette. It remembers your preference, and text stays in a comfy ~760px reading column instead of sprawling edge-to-edge.

🐛 Fixed

  • PDF text extraction got smarter. 📄 Text hiding inside PDF Form XObjects now gets pulled out, and oversized horizontal jumps are correctly read as column gaps instead of glued-together words. Your Cyrillic invoices thank you.

🧹 Housekeeping (aka the diet)

  • The installer lost 22% of its weight. 🪶 Swapped the chunky pdf-extract dependency for a tiny custom extractor on lopdf, which unblocked panic = "abort" and shed ~6 MB of unwind tables. Results:

    • NSIS installer: 5.06 → 3.94 MiB (−22%)

    • skim.exe: 17.2 → 11.2 MiB

  • The new extractor also never panics on malformed PDFs and has a decompression-bomb guard, because trust issues are healthy.

  • Trimmed unused tokio / tracing features while we were in there. Less code, less to maintain — the Skim way.

👉 Get it here:

40 views

Add a comment

Replies

Best

Nice update! Supporting any OpenAI-compatible endpoint is a great addition, especially for people running Ollama or other local models. These kinds of improvements usually have a bigger impact than flashy new features :)

 respecting local LLMs 🫡

the bring-your-own-endpoint thing is great for the homelab crowd. does Skim auto-detect context window size per model, or do you have to manually tell it the limits for something like a local Ollama model so it doesn't just truncate silently on a long PDF?

 Honest answer: right now it does neither — no per-model context probing, no field to set a limit. Skim just caps attachment/PDF text to a fixed, conservative budget before the request leaves your machine, and marks anything trimmed with a visible …(truncated) — safe, but it does leave headroom on a big local model.

Fair nudge, so I opened an issue to make it adapt (for Ollama the model's context length is basically already available to us):

Thanks! 🙏