Guide · Agent skill
How to use the Understand Dashboard Launcher agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The understand-dashboard skill is an agent skill that starts a local, interactive web dashboard showing the knowledge graph of a codebase that has already been analyzed by the companion /understand command. It is meant for developers who want to browse a project's structure and relationships visually, and for teams that want a local map of an unfamiliar codebase. By the end of this guide you will know what the skill does, how to install it, how to ask for it in plain language, and what to check before running it.
Quick start
- 1Generate the knowledge graph. Run the /understand command on your project first. The dashboard only shows an existing knowledge-graph.json file, so this step must come before anything else.
- 2Download and unzip the skill. Download the understand-dashboard ZIP from this page and unzip it into ~/.claude/skills/ for a personal install or into the project's .claude/skills/ folder for a project install.
- 3Or upload it in the Claude apps. In the Claude apps, open Settings, go to Capabilities, then Skills, and upload the skill.
- 4Ask for the dashboard. In your agent, ask to open the dashboard for the project, or type /understand-dashboard with an optional project path.
- 5Open the tokenized link. Copy the full Dashboard URL that includes ?token= from the agent's reply and open it in your browser.
- 6Stop the server when finished. Go to the terminal where the server is running and press Ctrl+C.
What the understand-dashboard skill does
An agent skill is a set of instructions, stored in a SKILL.md file, that an AI agent such as Claude can read and follow when your request matches the skill. This one is named understand-dashboard and is described as a way to launch the interactive web dashboard to visualize a codebase's knowledge graph.
When it runs, the agent works through a fixed sequence. It finds the project directory, checks that a knowledge-graph.json file exists, finds the dashboard code, starts a local server, and gives you a link to open in your browser. The dashboard only displays analysis that already exists; it does not analyze your code itself.
Check the requirements before you install
The skill needs a few things on your machine and in your project. The agent runs shell commands, so it needs Bash shell access. It also needs Node.js and npx. If the fallback build path is used, pnpm is required too.
Most importantly, the project must already have a knowledge-graph.json file produced by the /understand skill. Without it, the skill stops and tells you to run /understand first. Network access to GitHub releases is needed only for the optional prebuilt viewer path.
- Bash shell access
- Node.js and npx
- pnpm, for the fallback build path only
- An existing knowledge-graph.json created by /understand
- Network access to GitHub releases, optional, for the prebuilt viewer
Install the skill from the ZIP
Download the ZIP for the understand-dashboard skill from this directory page and unzip it. You can install it in one of two places. For a personal install that works across your projects, put the unzipped skill folder in ~/.claude/skills/. For a single project, put it in that project's .claude/skills/ folder.
If you use the Claude apps instead of Claude Code, you can upload the skill in the app under Settings → Capabilities → Skills.
The skill looks for the dashboard code in the Understand Anything plugin. It checks several locations in order, including ${CLAUDE_PLUGIN_ROOT}, ~/.understand-anything-plugin/, and a few clone-based install folders. If none of them contains packages/dashboard, the skill stops with an error that lists every path it checked. In that case, follow the installation instructions for the Understand Anything plugin on your platform.
- Personal install: unzip into ~/.claude/skills/
- Project install: unzip into the project's .claude/skills/
- Claude apps: Settings → Capabilities → Skills, then upload
Launch the dashboard with a natural request
You do not need to memorize a command. Once the skill is installed, ask the agent in plain language, for example that you want to see the project's knowledge graph in the browser. The skill's own slash name is /understand-dashboard, and it accepts an optional project path as an argument.
The skill uses the current working directory unless you give it a path. If you ask about a different project, name its folder so the agent points the dashboard at the right knowledge graph.
- Open the dashboard for this project
- Show me the architecture graph of ~/work/my-api
- Launch the understand dashboard so I can browse the file relationships
/understand-dashboard [project-path]What happens when the dashboard starts
The skill first tries a prebuilt viewer from the version-matched GitHub release. This path needs no install and no build. If that fails because there is no release asset for the installed version or there is no network, the skill falls back to installing dependencies with pnpm, building the core package, and starting a Vite dev server on 127.0.0.1.
Either way, the server prints a line with a tokenized link. The dashboard requires that token, and without it you will see an Access Token Required gate. The agent should give you the full URL including ?token=. The dashboard opens in your default browser automatically, and if port 5173 is busy, the next free port is used. To stop it, press Ctrl+C in the terminal where it runs.
- Prebuilt viewer is tried first and is pinned to the installed plugin version
- Fallback uses pnpm install, a core build, and the Vite dev server
- The link must include the ?token= value
- Port 5173 is used by default, or the next available port
npx --yes "$VIEWER_URL" "$PROJECT_DIR"
cd "$DASHBOARD_DIR" && GRAPH_DIR="$PROJECT_DIR" npx vite --host 127.0.0.1Safety notes: scripts, network and shared machines
This skill runs scripts and uses the network. It executes shell commands and may download a viewer tarball from GitHub releases. The server is started locally and runs in the background, so it keeps running after you stop reading. The token in the URL is what protects the knowledge graph data, so anyone who obtains that full link could open the dashboard while the server runs.
For these reasons, do not use the skill on a shared or untrusted machine. Read the SKILL.md file before installing if you want to see every command the agent will run.
- Runs shell commands and background servers on your machine
- May download a prebuilt viewer from GitHub releases
- Do not paste the tokenized link where others can see it
- Stop the server with Ctrl+C when you are finished
What people use it for
- Get a visual overview of an unfamiliar codebase before making changes to it
- Explore how files and modules relate to each other after running /understand
- Share a local, browsable map of a project with teammates on a trusted network, using the link carefully
- Revisit the architecture of a project you have not worked on for a while
Frequently asked questions
How do I install the understand-dashboard skill?
Download the ZIP from this page and unzip it into ~/.claude/skills/ for a personal install or into a project's .claude/skills/ folder. In the Claude apps, you can instead upload it under Settings → Capabilities → Skills.
What do I need to run the understand dashboard?
You need Bash shell access, Node.js and npx, and an existing knowledge-graph.json created by /understand. pnpm is needed only for the fallback build path, and network access to GitHub releases is optional.
Why does the skill say No knowledge graph found?
The skill only displays existing analysis. Run /understand on the project first so that a knowledge-graph.json file exists, then ask for the dashboard again.
Why does the dashboard show an Access Token Required screen?
The dashboard needs the token that is printed in the server output. Use the full link that includes ?token= rather than just the base localhost address.
Is the understand-dashboard skill free to use?
The project is listed under the MIT licence. Check the licence file in the source repository for the full terms.
How is this different from other ways to view a codebase graph?
This page does not compare it with other tools. The skill's job is narrower: it starts a local dashboard for a knowledge graph that /understand has already produced, and it runs entirely from the agent's commands on your own machine.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use this skill if you already run /understand on a project and want to browse its architecture in your browser on your own machine. Pick something else if you have not generated a knowledge graph yet, if you need a hosted or shared service, or if you work on a shared or untrusted computer where a locally running server with a token is a risk.
Download the skill