Production readyAIv1.0.0 · updated Jul 1, 2026
AI

Open Source Agent Workbench Template

A workspace for prompts, tasks, artifacts, and coding-agent runs.

An AI-native foundation for managing agent runs, prompt recipes, generated artifacts, and review loops.

Runnable Agent-ready Install-ready Deploy-ready
foundationbase.app/demo/agent-workbenchOpen full demo →

This is the real app, not a mockup. Whatever you enter in the demo stays in your browser and belongs to nobody.

What you get

One zip. Everything to own it.

Agent Workbench is built for the AI coding era: a practical foundation for teams that want to capture prompts, run context, generated files, review notes, and follow-up tasks. It gives agents a predictable domain model while giving humans a clean audit trail.

  • Agent run timelines
  • Prompt and context libraries
  • Generated artifact review
  • Human approval states
  • Provider-agnostic model boundaries
agent-workbench.zipMIT
agent-workbench/├── app/the running Next.js app├── agent/START_HERE, PROJECT_MAP, RULES, RECIPES, TESTS, DEPLOY├── CLAUDE.mdread automatically by Claude Code├── AGENTS.mdread automatically by Codex├── .cursor/rules/loaded automatically by Cursor└── README.mdrun it, then make it yours

Inside the app

What the running app actually looks like
Agent Workbench: Agent runs
Agent runsEvery run with its prompt, provider, and human approval state.

Customize with AI

Your agent already knows this codebase.

The repository is organized around explicit agent concepts and ships with recipes for adding providers, artifacts, and review workflows.

CodexClaude CodeCursor

Prompts written against this app

Change the branding, copy, and layout for my Agent Workbench app.
Add or remove fields so it matches my workflow.
Add export and import for my ai data.
Add optional sync later without changing the local-first default.

Run it

5 min to your own copy.

  1. 01Create your editable app copy and unzip it.
  2. 02Run npm install.
  3. 03Run npm run dev and open http://localhost:3000.
  4. 04Read /agent/START_HERE.md before customizing.
  5. 05Swap the local-first storage boundary for your own backend when ready.
Full walkthrough, including deploy and install →
Where your data lives

Agent Workbench data stays in this browser by default.

No account, backend, or cloud sync is included in the current package.

An agent can add optional sync later while preserving the local-first default.

Honest limitation

This package is standard-ready for Create My App and AI customization.

A full app-only PWA install route is not enabled yet for this app.

Verification receiptchecked Aug 2, 2026 at c11c184 · v1.0.0
Demo renderscaptured 2026-07-26, 28 KBpassed
Package installs and buildsinstalled, typechecked and built in 63s on node v24.13.0passed
Source package complete31 files, 48 KBpassed
Agent instructions included5 handoff files presentpassed
No network calls in app sourcestatic scan clean: no network calls, eval, or secrets in app sourcepassed
License matches the packageMIT, matching the file in the packagepassed

If this isn’t quite right

All AI foundations →

Take Agent Workbench and make it yours.

Free, MIT, 5 min to running. The demo above is exactly what lands in your folder.