| What it does | Multi-model desktop workspace with cost optimization through agent routing
| Local-first AI coding memory and recovery for Claude, Cursor, Codex, Gemini, and more
|
|---|
| Stage | Not given
| Not given
|
|---|
| Founded | Not given
| Not given
|
|---|
| Based in | Not given
| Not given
|
|---|
| Pricing model | Freemium
| Paid
|
|---|
| Pricing | Desktop app: Free and open source (Apache 2.0). Halv Cloud: US$10 per month.
| $49 one-time lifetime license, no subscription or recurring fees, Windows/macOS/Linux desktop apps included
|
|---|
| Key features | - JEV agent routing for automatic model and reasoning selection
- Support for Claude Code, Codex, Kimi, and GLM models
- Synced savings dashboard
- History across machines
- Automatic account rotation
- Code navigation via Crux
- Command output filtering via RTK
- Model context processing via Headroom
- Cross-platform support for macOS, Windows, Linux
| - Session recovery across 31 AI coding tools including Claude, Cursor, Codex, Gemini, Cline, Devin, Windsurf, and more
- Git-aware state tracking with branch, diff, changed files, and recent commits
- Dependency graph showing file relationships and god-node detection
- 49 MCP tools callable directly from Claude Code, Cursor, Devin, and Windsurf
- Session Vault for local recovery history after crashes and outages
- Zero telemetry with local recovery and optional explicit exports only
- Hivy on-device intelligence for context trimming and drift detection
- Windows, macOS, and Linux desktop apps with CLI and VS Code/Cursor extension
- Council 12-agent review layer for risky handoffs to surface stale assumptions
|
|---|
| What makes it different | - 57.1% lower recorded model costs compared to vanilla Codex on benchmark runs
- Free and open source desktop app under Apache 2.0 license
- Agent routing included with Halv Cloud at no additional fee
- Single workspace for multiple AI coding models
- Matched vanilla Codex pass rates with lower cost
| - Recovers from disk state without needing the old session alive, unlike transcript-based approaches
- Works after the session ends or is locked out, unlike /compact or /handoff skills
- Formats for 31 AI tools instead of locking you into one platform
- Local-first with zero telemetry and no cloud sync unless explicitly configured
|
|---|
| Free plan | Not given
| Not given
|
|---|
| Open source | Not given
| Not given
|
|---|
| Platforms | Not given
| Not given
|
|---|
| Public API | Not given
| Not given
|
|---|
| Community votes | 0 | 0 |
|---|
| DR (Domain Rating by Ahrefs) | 19 ● no change since last check | 8 ● no change since last check |
|---|
| Trust Flow (Majestic) | 3 | 5 |
|---|
| Citation Flow (Majestic) | 24 | 20 |
|---|
| Referring domains (Majestic) | 57 | 24 |
|---|
Halv
ShardStitch