Bring your own key
OpenAI, Anthropic, Gemini, OpenRouter, xAI, Mistral, Groq — plug in whatever you already pay for. Verify the key in-app, then use the same quota everywhere.
SinWeave
v1.0.0-α
Download
SinWeave
v1.0.0-α
Download
SinWeave is a free, AI-native editor built on VS Code. Bring your own API keys, run the best models — or your own, locally — and keep your code on your machine. No middleman, no lock-in, no monthly seat.
SessionStore, keep the existing error shapes.
auth.ts, sessions.ts, and 3 callers. I'll
preserve AuthError and switch
getToken() to SessionStore.read().
42- const token = await legacyTokens.get(userId) 42+ const token = await SessionStore.read(userId) 55 if (!token) throw new AuthError('expired')
SinWeave puts every layer of the AI stack back in your hands — and keeps everything you already love about VS Code.
OpenAI, Anthropic, Gemini, OpenRouter, xAI, Mistral, Groq — plug in whatever you already pay for. Verify the key in-app, then use the same quota everywhere.
Ollama, LM Studio, vLLM, and any OpenAI-compatible endpoint auto-detect. Run everything on your own hardware — your code never leaves the machine.
Ask for a change across the repo. SinWeave reads, reasons, and writes coherent diffs you apply or reject one hunk at a time — never a stealth write.
Select, invoke, describe. A focused inline editor rewrites the selection right where your cursor is — no modal dance, no context loss.
Tab completion backed by your model of choice — aware of the file, the project, and your recent edits. Swap providers without switching editors.
SinWeave is a true VS Code fork — your extensions, keybindings, themes and settings transfer in as-is. Open it and keep shipping; no muscle-memory tax.
From download to your first AI edit in about two minutes. No account, no onboarding wizard.
Download, drag to Applications, open your folder. Your extensions, themes and keybindings are already there — it's the editor you know.
Drop in an API key from any provider and verify it in-app — or let SinWeave auto-detect Ollama, LM Studio, or vLLM running on your machine.
Chat across files or hit ⌘K inline. Every change lands as a reviewable diff — accept, reject, or edit each hunk. Your Git history stays clean.
$ sinweave . → starting local agent… → indexed 1,248 files · 84 ms → model: openrouter/claude-sonnet (usd 0.87 used of 10.00) ⌘K refactor auth.ts to use SessionStore ✓ 4 files changed, 38 +/- 12 · ready to review $
Cursor and the rest route your code through their servers and meter you monthly. SinWeave doesn't sit between you and the model at all.
Direct connections to whatever you already use. No middle tier.
Available for macOS · more platforms on the way
Signed & notarized. Requires macOS 11 or later · 8 GB RAM recommended.
Yes. The editor itself is free — no seats, no tiers, no trial timers. If you bring an API key, you pay the model provider directly (not us). If you run local models with Ollama, LM Studio, or vLLM, you pay nothing at all.
No. API requests go straight from your editor to whichever provider you configured. There's no SinWeave server between you and the model. If you use only local models, nothing leaves your machine — ever.
No middleman, no required account, no metered plan, no mandatory cloud tier. You pick the models you want — frontier, budget, or fully local — and SinWeave wires them in directly. You keep the keys, the quotas, and the code.
Yes. The extensions, keybindings, themes, and settings you already rely on transfer in as-is. Open SinWeave and everything's where you left it.
Packaged releases ship with a signed, notarized auto-updater. New versions install in the background when available; you can pause or disable updates anytime in Settings.
Yes. SinWeave is free for individuals and teams — no per-seat fees. The model provider you plug in may have its own terms, so check with them if you're shipping on their API in production.