Get started
Welcome to Faundr
Faundr is the dashboard for people who build software with an AI agent, even if they don't write code. It follows what the agent does in your project, remembers what was decided, checks security, quality, tests and design, and shows all of it in plain language in one place.
It solves three problems you run into when an AI writes your code:
- The AI forgets. Every new conversation with the agent starts from scratch. At the start of each session, Faundr hands it where the work stopped, the team's rules, the decisions already made and what's still open.
- You can't see what's happening. Code changes fast and you can't read all of it. Faundr shows what's being built, what broke, what became unsafe and what was left half done.
- One mistake can be expensive. A leaked key, a wiped database, a fake package. Faundr blocks what can't be undone, asks for your confirmation on risky actions and saves restore points so you can roll back.
The three parts
Faundr has three parts that work together.
| Part | Where it lives | What it does |
|---|---|---|
| The dashboard | This site | Shows everything Faundr knows about the project: features, memory, activity, findings, tests, errors and cost. This is where you and your team follow the work. |
| The plugin | Inside your AI agent (Claude Code or Codex), on your computer | Watches sessions, hands context to the agent, runs the checks, guards against disasters and sends results to the dashboard. It also adds the /faundr:… commands. |
The faundr CLI | On your computer, installed with the plugin | The terminal commands the plugin and the agent use. You can run them yourself too. |
The plugin does the work and the dashboard shows it. Until the plugin is installed and a folder is linked, the dashboard stays empty.
What runs on its own and what you ask for
Most of Faundr works without you doing anything. Once the plugin is installed and the folder is linked to a project:
- every conversation with the agent becomes a session in the dashboard's log;
- the agent gets the project's memory at the start of each session;
- security, quality and design are checked automatically on your computer, without AI;
- errors in the agent's commands become items to fix, and they close on their own once the command passes again;
- dangerous commands are blocked or need your confirmation;
- a restore point is saved before changes;
- at the end of each response, the agent is reminded to prove its work and to leave a handoff note about what's still pending;
- the skills you already use get linked to the project: the agent is reminded to use one when a request touches its topic, and the ones that find problems (security, tests, quality or errors) send what they find to the dashboard. See Workflows.
Other things you ask for with a command in your agent. For example, /faundr:security runs an AI security review, /faundr:launch tells you whether you're ready to launch, and /faundr:undo takes you back to when things worked.
See everything that runs automatically in What Faundr does on its own.
Where to start
- Install the plugin and link your first project. It takes about 5 minutes.
- Read A day with Faundr to get a feel for the rhythm.
- When a term trips you up (feature, rule, restore point, S-3, P-1…), check Concepts and vocabulary.
Your privacy
The security, quality, design and test checks run on your computer. Only the results go to the dashboard: the file name, the line and a short excerpt, with keys masked. The session log is different. It stores your requests and what the agent did at each step, which includes excerpts of code it read or edited, with keys replaced by ***. Only you can see that conversation; your team sees the summary. Find out exactly what leaves and what stays in What leaves your computer.