Releases: bdeb-technology/bdebtech-ai-ide-releases
Releases · bdeb-technology/bdebtech-ai-ide-releases
Release list
BdebTech AI IDE 0.2.0
BdebTech AI IDE 0.2.0 - repository awareness, Studio checkout, hardening
Downloads
- Windows x64 installer:
BdebTech-AI-IDE-Setup-0.2.0-win-x64.exe(VSCodium 1.135 + BdebTech AI extension + bdebd daemon). No admin rights needed. Not code-signed yet: SmartScreen ? More info ? Run anyway; verify withSHA256SUMS.txt. - Extension only:
bdebtech-ai-ide-0.2.0.vsixfor VS Code / VSCodium / Cursor (needs the daemon). - Daemon only:
bdebd-0.2.0-win-x64.exe- OpenAI-compatible local API on127.0.0.1:41337(token in%LOCALAPPDATA%\BdebTech\AI\token).
What's new
- Repository awareness: index a workspace locally (Nomic Embed Text v1.5 on CPU, SQLite). Chat gets an
@codebasetoggle; autocomplete receives related snippets from other files via repo-level fill-in-the-middle, so completions use your real functions and types. - Multi-file agent (
Ctrl+Alt+Shift+K): schema-constrained file selection and full-file edits with a reviewable diff per file (Lite+ or Studio models). - Custom GGUF models:
bdebd models add <path|url>or "Add a custom GGUF?" in the switcher. - Studio trial and checkout: "Start Free Studio Trial" (7 days) inside the IDE; buy with UPI / USDT / Binance Pay at https://ide.bdebtech.in/buy/.
- Security: per-install API token on the daemon, CORS limited to editor webviews.
- Updates: daily check of ide.bdebtech.in/releases.json with an in-editor prompt.
- Export Diagnostics command for support (secrets redacted).
- Autocomplete stops at the end of the enclosing block and never includes markdown chatter; model switching during engine autostart fixed; smarter engine selection for integrated graphics.
Upgrading from 0.1.0: run the new installer over the old one. Models and settings are kept. Other tools that use the local API must now send the token as their API key.
Docs: https://ide.bdebtech.in/docs/ ? Changelog: https://ide.bdebtech.in/changelog/
BdebTech AI IDE 0.1.0 - public preview
BdebTech AI IDE 0.1.0 - first public preview
Offline-first AI code editor. Local open-weight models, zero cloud, your code never leaves your PC.
Downloads
- Windows x64 installer:
BdebTech-AI-IDE-Setup-0.1.0-win-x64.exe(VSCodium 1.135 + BdebTech AI extension + bdebd daemon). No admin rights needed. - Extension only:
bdebtech-ai-ide-0.1.0.vsixfor VS Code / VSCodium / Cursor (needs the daemon). - Daemon only:
bdebd-0.1.0-win-x64.exe- OpenAI-compatible local API on127.0.0.1:41337.
On first launch the IDE downloads the Starter model (Qwen2.5 Coder 0.5B, 645 MB) and the matching llama.cpp engine. After that it works fully offline.
What's new
- Tab autocomplete (fill-in-the-middle) with single-line / multi-line heuristics
- Sidebar chat with file and selection context; copy / insert / apply code actions
- Edit with AI: select, describe, review diff, apply
- Starter 0.5B (CPU-first) and Lite+ 1.5B free models; Studio 7B / 14B / 32B behind license
- Hardware profiling with CUDA / Vulkan / Metal / CPU engine selection and automatic fallback
- Resumable, checksum-verified downloads; offline Ed25519 license verification with 30-day grace
Known limitations
- Binaries are not code-signed yet: Windows SmartScreen will warn. Verify with
SHA256SUMS.txt. - Linux and macOS installers follow in a later release.
- No repository-wide retrieval yet; chat context is what you attach.
Measured on a 2021 Intel i5 laptop, CPU only: ~50 tokens/s, first token ~150 ms.
Docs: https://ide.bdebtech.in/docs/ · Pricing: https://ide.bdebtech.in/pricing/