AgenticBox — Vercel for AI agents. Local-first, vendor-neutral sandbox runtime for autonomous AI agents with browser automation, filesystem tools, and terminal access.
Resources
13Install
npx skillscat add morpheus-sh/agenticbox Install via the SkillsCat registry.
SKILL.md
AgenticBox Company Brain
Identity
- Domain: agenticbox.co
- Positioning: "Vercel for AI agents" — deploy agents without worrying about sandboxes, permissions, browser sessions, secret management, observability, cost controls
- Target audience: Developers and companies deploying AI agents to production
- Tech stack: Rust (daemon + crates), Tauri v2 desktop, Python FastAPI agent runtime, Docker/Rancher Desktop, SQLite
- License: MIT OR Apache-2.0
Pricing Model
| Tier | Price | Target |
|---|---|---|
| Starter | $49/mo | Solo devs, small projects |
| Pro | $199/mo | Teams, production workloads |
| Enterprise | $999/mo | Large orgs, custom SLAs |
Key Features (by phase)
- Phase 1 (Ready): Container sandbox lifecycle, terminal tool, filesystem mounts + guard, OpenAI-compatible API, Tauri desktop app, session persistence
- Phase 2 (Planned): Browser automation (Playwright), permission system UI
- Phase 3 (Planned): Firecracker microVMs, policy engine
Workflow
- Load this skill when working on AgenticBox
- Check
docs/ARCHITECTURE.mdfor component details - Use
cargo run --bin daemonfor daemon-only dev - Use
cd apps/desktop && pnpm tauri devfor desktop dev - Run
./scripts/dev.shfor full stack
Marketing Notes
- Emphasize "local-first, vendor-neutral" — no cloud lock-in
- Highlight Rust performance + Tauri desktop experience
- Target: developers who are tired of managing agent infrastructure
- Content strategy: blog posts on agent deployment patterns, sandbox security, benchmark comparisons
Supermemory Container Tag
Use containerTag: "agenticbox" when storing memories about this project.