v0.1.0

First release of Agent Dashboard: a local, single-user web UI that runs Trustabl's Analyzer, Forge and Probe on your own repos. It listens on 127.0.0.1 only.

git clone --branch v0.1.0 https://github.com/trustabl/agent-dashboard.git
cd agent-dashboard
make run            # then open http://127.0.0.1:4317

Requirements: macOS, Linux or WSL2, with git, curl, tar and make. If Node.js 22.13 or newer isn't installed, make run installs a private copy. Probe also needs Docker (or OpenShell) and an OpenAI-compatible LLM endpoint.

What's in it

• Overview, Forge, Analyzer and Probe pages. Generate skills (trustabl forge), run reliability scans (trustabl scan) and network access probes (autofix fix).
  • Output streams live, runs keep going when you close the tab, and you can cancel them.
  • Every run has Copy CLI, which copies the exact command.
• Pinned CLIs installed for you.
  • make run downloads trustabl 0.1.15 and autofix 0.5.0 from their GitHub releases and checks their SHA-256 against tools.lock.json.
  • Only missing versions are downloaded, so a restart downloads nothing.
  • Offline, it warns and still starts, and each CLI installs on its first run.
• Switch CLI versions. From the Forge, Analyzer and Probe pages, install the latest release (checked against that release's checksums.txt), switch between installed versions, or go back to the pinned one.
• Local state only. Everything lives under ~/.agent-dashboard, and make wipe deletes it. The dashboard never stores API keys. Probe sends tool source only to the LLM endpoint you configure.

Tested

CI runs these on every pull request and on main:

• lint, typecheck and unit tests;
• Playwright end-to-end tests with WCAG 2.1 AA checks;
• artifact parity (a dashboard run gives the same artifacts as the same CLI run);
• a gitleaks secret scan;
• make run on a clean macOS and a clean Ubuntu.