Why it matters
This dashboard offers AI builders a streamlined local environment to assess agent reliability and network behavior without needing to install individual tools. It simplifies the testing process, allowing for quicker iteration and debugging of AI agents.

What changed

Trustabl has introduced the Agent Dashboard, a new, single-user, local web interface designed to simplify the process of evaluating AI agents. This dashboard consolidates the functionality of Trustabl's existing open-source tools: Analyzer, Forge, and Probe. Users can now run reliability scans, test agent skills, and perform network access probes directly from this unified UI, eliminating the need to install and manage each component separately. The latest release, v0.1.0, provides immediate access to these capabilities. The dashboard operates exclusively on 127.0.0.1, ensuring that no user or organizational data is managed centrally, and there is no login requirement. It pins specific CLI versions, trustabl 0.1.16 and autofix 0.5.0, which are downloaded and verified via SHA-256 checksums during the make run process. Newer versions of these CLIs can be installed directly from the dashboard's Forge, Analyzer, and Probe pages. Crucially, user source code remains on the local machine; only tool source is sent to a configured LLM endpoint when using the Probe feature.

Why it matters for builders

For AI builders, the Agent Dashboard significantly lowers the barrier to entry for robust agent testing. It provides a consolidated, user-friendly interface that abstracts away the complexities of setting up and integrating multiple specialized tools. This means developers can more efficiently assess the reliability, functionality, and network interactions of their agents, leading to faster development cycles and more stable AI applications. The local-first approach also enhances security and privacy, as sensitive code and configurations do not need to be uploaded to external services.

Practical impact

To get started, developers need macOS, Linux, or WSL2 with git, curl, tar, and make installed, along with Node.js version 22.13 or newer. The make run command will handle the installation of Node.js if necessary. For the Probe functionality, Docker (or OpenShell) and an OpenAI-compatible LLM endpoint, such as Ollama, are also required. The setup involves cloning the repository, navigating to the directory, and executing make run or ./scripts/run.sh. Once the dashboard is running at http://127.0.0.1:4317, users can initiate scans, generate skills, or run probes by specifying a local agent path. The output streams live, and runs can be managed even if the browser tab is closed. For local LLM testing with Probe, users can set up Ollama by serving the model and pulling a suitable image like qwen2.5-coder:32b, then configuring the dashboard with the appropriate model name and LLM URL. Cloud LLM access is managed via environment variables like OPENAI_API_KEY or ANTHROPIC_API_KEY, with the dashboard never storing these keys.

State management is handled within ~/.agent-dashboard, with console.db storing repository data, run history, and findings, and the runs/ directory containing artifacts from each execution. A make wipe command is available to clear all stored state and downloaded CLIs. Development workflows are supported with make dev for hot reloading, make check for linting and type checking, make e2e for end-to-end testing, and make parity to verify CLI artifact consistency.

Caveats and source limits

This release is described as a "fresh release" with version v0.1.0, indicating it is likely an early iteration. The source mentions "Policy and telemetry via OpenTelemetry are coming soon," suggesting these features are not yet implemented. While the dashboard supports local LLM endpoints like Ollama and cloud models via API keys, specific performance benchmarks or comparisons between different LLM providers are not included in the provided information. The dashboard's primary focus is on local execution and testing, and it does not appear to offer features for distributed agent deployment or management. The excerpt also notes "6 AI signals, 6 developer signals," but the specifics of these signals and their interpretation are not detailed.

Sources

Written with AI assistance from the linked sources; every claim below was checked against them automatically. How we produce articles.

Claim check: 9/9 supported claims - 9 evidence links - 100% avg confidence
  • Agent Dashboard runs reliability scans, skills, and network access probes on AI agents.supported - github.com
  • Agent Dashboard integrates Trustabl's Analyzer, Forge, and Probe tools into a single local web UI.supported - github.com
  • Agent Dashboard runs on 127.0.0.1 only, with no login, user, or organization model.supported - github.com
  • Agent Dashboard requires macOS, Linux, or WSL2, Git, Curl, Tar, Make, and Node.js 22.13 or newer.supported - github.com
  • Agent Dashboard requires Docker (or OpenShell) and an OpenAI-compatible LLM endpoint for Probe functionality.supported - github.com
  • Agent Dashboard stores state in ~/.agent-dashboard, including console.db for runs and findings.supported - github.com
  • Agent Dashboard is licensed under Apache-2.0.supported - github.com
  • Agent Dashboard has 62 stars and 51 forks on GitHub.supported - github.com
  • Agent Dashboard is a fresh release, version v0.1.0.supported - github.com

Caveats

  • This is a GitHub repository description.
  • GitHub metrics can change rapidly.
  • Single-source caution: verify critical details at the linked source.
Radar score 86/100 - how it was calculated
Reliability82
Freshness92
Novelty73
Technical85
Developer96
Ecosystem72
Confidence96
  • Reliability 82: GitHub metadata supports source trust
  • Freshness 92: Fresh GitHub activity
  • Novelty 73: Fresh GitHub release
  • Technical 85: Repository technical metadata
  • Developer 96: Developer tooling signals
  • Ecosystem 72: Fresh GitHub release
  • Confidence 96: Claims have reliable evidence
Share
XLinkedInHacker News

Discussion

Loading comments...

Related articles

Agents - Nov 17, 2026BuilderIO Releases Agent-Native TypeScript FrameworkBuilderIO has released Agent-Native, an open-source TypeScript framework for building agentic applications. It enables agents to pair autonomous work with purpose-built UIs, allowing shared actions, data, and application state between the agent and the user interface.Agents - Nov 17, 2026Overleaf CLI (olcli) v0.17.0 Adds AI Agent IntegrationThe Overleaf CLI tool, olcli, has released version 0.17.0, introducing an MCP server for AI agents. This allows AI agents to interact with Overleaf projects, prompting users before making changes.Agents - Nov 16, 2026Node9 Proxy Adds Access Control for AI AgentsNode9 Proxy has released version 2.61.1, introducing enhanced access control for AI agents. This tool allows developers to define permissions for agents interacting with tools like Claude Code, Codex, and Gemini, and provides auditing capabilities.Agents - Nov 16, 2026aisync: Context Management for AI Coding Agentsaisync is a new local hub designed to manage context and coordinate multiple AI coding agents. It offers features like shared rules, a hash-chained journal for verification, file claims to prevent conflicts, and mid-session updates for agents like Claude Code, Codex, and Cursor.Agents - Nov 14, 2026Lich v0.90.0: Multi-Agent IDE with Worktrees and Browser ControlLich has released version 0.90.0, a multi-agent IDE that allows multiple AI agents to work concurrently on separate branches within a single window. The tool supports agents like Claude Code and Codex, managing their tasks in individual worktrees and offering features like inline code review and browser-driven agent interaction.Agents - Nov 17, 2026ESP32 iPhone Agent enables AI control of iPhonesThe ESP32 iPhone Agent project allows AI models to control an iPhone using a low-cost ESP32 microcontroller. It enables AI to see the screen via AirPlay, interact using USB HID for touch input, and process audio for voice commands, all without requiring a Mac, developer account, or jailbreaking the device.