Local-first AI Agent control workbench

Operate OpenClaw from one workspace.

Dashboard, files, IDE, model routing, channels, CLI agents, and system recovery — packaged as one maintainable extension.

Open sourceLocal-firstCredentials stay localMaintenance mode
Tracevane dark operations dashboard showing agents, model routes, runtime health, and recent events
ILLUSTRATED PRODUCT PREVIEWTRACEVANE / LOCAL
01 / Capabilities

Everything around OpenClaw, connected.

Tracevane keeps each runtime boundary visible while bringing daily operations into one coherent surface.

01

Dashboard

Unified health, routing, channels, Agent runs, and recovery status.

02

Workspace & IDE

Files, online editing, preview, terminal, Git, and a project workbench.

03

Model Gateway

Route models and providers through explicit protocol and health boundaries.

04

Channels & CLI Agents

Connect message inputs to Codex, Claude Code, and OpenCode workflows.

05

System Guard

Inspect configuration, device trust, service health, and recovery evidence.

06

Standalone / Gateway

Use a local 3760 entry or mount Tracevane at the OpenClaw Gateway.

Tracevane Gateway architecture connecting CLI, HTTP, WebSocket, MCP, system guard, workspace, and model providers
TRACEVANE GATEWAY / EXPLICIT INPUT AND PROVIDER BOUNDARIES
02 / Install

Inspect first. Then install.

OpenClaw must already be installed and onboarded. The Tracevane installer verifies Release metadata and SHA-256.

Linux, macOS, or WSL2 · initialized OpenClaw requiredNo OpenClaw yet? Start with the fresh-machine guide →
03 / Open source

Documentation & community

Clear install paths, public maintenance boundaries, and direct links to the source.

Built for self-hosters and contributors.

Tracevane is open source and in maintenance mode. Contributions that improve stability, documentation, and compatibility are welcome.

Read CONTRIBUTING.md

Join the project on GitHub.

Report reproducible issues, propose focused changes, and follow current maintenance work.

View on GitHub