Void: Open-Source AI-First Editor Forked From VS Code
Void (~29k stars) is an open-source, AI-first editor forked from the VS Code lineage — aimed at developers who want Cursor-like AI workflows with more transparency, fewer black boxes, and control over where prompts go. It is not “VS Code with an extension” — it is a full editor distribution with AI woven into the core UX: chat, composer-style edits, inline assistance, and model routing you can inspect and tune.
Who considers Void
- Teams evaluating open AI editors before standardizing on a commercial fork
- Developers sensitive to telemetry and vendor lock-in who still want integrated agents
- Power users comfortable living on fast-moving forks for daily driver work
- Privacy-conscious orgs pairing Void with local or VPC-hosted models
- Cursor migrants who want familiar UX without subscription coupling
Void vs Cursor vs stock VS Code + Cline
Cursor ships polish, proprietary orchestration, and tight model deals — excellent velocity, opaque internals. VS Code + Cline keeps the stock editor and bolts on an agent — minimal fork risk, two-vendor feel (Microsoft + plugin). Void bets that an open fork can merge AI into the shell like Cursor does while staying inspectable on GitHub.
None replace judgment: all three need rules, tests, and review. Void’s bet is governance-friendly teams will accept fork churn for source-available behavior.
Installation and updates
Download builds from Void releases or compile from source per README. Forks lag upstream VS Code security patches — assign an owner to track releases.
git clone https://github.com/voideditor/void.git
cd void
# Follow README: yarn/npm install, compile, package
# macOS / Windows / Linux artifacts differFirst launch: connect API keys or local inference endpoints, import your VS Code settings/extensions where compatible, copy .cursorrules or project rules from awesome-cursorrules.
Day-to-day workflow
Inline edits: Select code, ask for refactor — same muscle memory as Cursor composer.
Multi-file features: Describe a ticket; Void proposes file diffs across the tree — review each hunk.
Model switching: Use fast models for small edits, frontier models for architecture — configure per task if the UI allows.
Rules-first: Commit stack rules so Void does not hallucinate dependencies (package verification).
Git discipline: Small commits, branch per experiment — AI editors make large diffs easy and large mistakes easier.
Extension compatibility reality check
VS Code forks break extensions occasionally — marketplace signing, API drift, or disabled Microsoft-only services. Pilot Void on a non-critical machine before mandating it org-wide. Keep a stock VS Code or JetBrains install for comparison. Test your non-negotiables: ESLint, Prettier, language servers, debugger, remote SSH.
Pairing with the rest of your toolchain
Void is the editor layer. Long-horizon autonomy may still belong in OpenHands. Terminal git workflows may still use Aider for scripted refactors. Browser builders like bolt.diy handle greenfield spikes you then open in Void. Study prompt patterns via system prompts research but write your own rules.
WordPress development
Void works for PHP/JS WordPress development if your extensions cooperate — same as VS Code. For Bedrock/Sage shops, encode Composer scripts and PHPCS in rules. Never use the AI shell to bulk-update post meta; editors should use Quickfields Bulk Editor for ACF for spreadsheet ACF workflows while developers stay in Void for theme/plugin code.
Risks of living on a fork
- Update lag behind Microsoft security fixes
- Extension breakage after upstream API changes
- Team friction if only some developers use Void
- Model costs unchanged — open editor ≠ free inference
- Over-trust of integrated AI because it feels “official” in the fork
FAQ
Is Void a drop-in Cursor replacement? UX is similar; extension parity and model features differ — pilot before switching.
Can I use corporate SSO models? Depends on Void’s provider support for your endpoint — verify in settings/docs.
Does Void phone home? Read the project privacy docs and network tab for your build — open source does not automatically mean zero telemetry.
What about Continue? Continue is a plugin; Void is the whole editor — some teams use neither overlap; some use Continue inside Void if supported.
Migrating from Cursor to Void
- Export Cursor keybindings and extensions list from stock VS Code compatibility layer
- Copy
.cursorrulesverbatim — Void reads the same convention files in most builds - Re-enter API keys; do not commit them to dotfiles in Git
- Run your top five daily tasks (debug, refactor, multi-file feature) as a smoke test
- Keep Cursor installed two weeks for parallel comparison before decommissioning
Expect friction on proprietary Cursor-only features; Void chases the open subset, not every commercial differentiator.
Telemetry and trust
Open source does not automatically mean zero phone-home. Read Void’s privacy documentation, inspect network traffic during a session, and document findings for your security team. Many developers choose Void specifically to pair with local Ollama or on-prem OpenAI-compatible gateways — the editor is only one hop in the data path.
When Void is worth the fork tax
Worth it: AI-native daily driver with inspectable code, policy against closed copilots, budget for keeping the fork updated. Not worth it: teams that only need occasional chat — Cline on stock VS Code is simpler operations. The ~29k stars signal interest, not guaranteed long-term maintenance — evaluate commit velocity quarterly.