browserbase/stagehand · Open source
Open-source AI browser automation framework from Browserbase that combines natural-language actions with deterministic browser control for testing, scraping, and agent workflows.

Preparing tokens&
Loading the next builder or enterprise surface.
Loading
Preparing page
Loading product graph, proof, and adoption context.
Loading⌘Developer workbench
Find products, backend paths, quickstarts, credits, perks, Build Packets, and project proof for the thing you want to ship.
Daily builder loop
The reason to join is not another directory account. It is a reusable build cockpit: stack decisions, perks, agent packets, backend choices, repo proof, and launch radar all stay connected to what you are actually shipping.
Stack context
Saved and reused across your next build.
Launch radar
Saved and reused across your next build.
Proof trail
Saved and reused across your next build.
Agent backend track
Tokens& compares partner-owned backend primitives, generates a backend-aware Build Packet, and records proof after the repo ships. We do not need to host your database to make the workflow valuable.
Data/auth
Postgres, auth, RLS, storage, vectors.
MCP handoff
Codex/Cursor context, env vars, evals.
Proof loop
Repo, demo, stack, usage, credits.
Supabase
Developer-native shipping
Publish from the place you already build. Install the Tokens& Agent Pack, then connect a project after login to generate scoped MCP config, paste-ready AI context, backend-aware Build Packets, and a launch checklist.
MCP config
Scoped tokens stay behind login.
Build Packet
Backend, docs, env vars, evals, and proof context.
Launch checklist
Repo, demo, stack, backend, and handoff links.
Try an example:
Launch radar
Featured launches, new products, and trending tools sit inside the workbench so discovery turns into saves, comparisons, and project proof.
New launches and quickstarts from the current build cycle.
browserbase/stagehand · Open source
Open-source AI browser automation framework from Browserbase that combines natural-language actions with deterministic browser control for testing, scraping, and agent workflows.
Fast publish
Paste a GitHub repo, add a demo URL, confirm the stack, and publish a portfolio page visitors can save, remix, and share.
GitHub repo
Readme, language, dependencies, and package scripts.
Demo URL
OpenGraph title, image, and reachability check.
Tools used
Stack items mapped to tokens& product profiles.
Build notes
What you shipped, what worked, and what visitors can remix.
Use the stack recommendation as your shortlist, then browse profiles for alternatives, docs, packaging, and proof.
Community opportunities
Find credits, perks, workshops, prizes, and community programs matched to what you are building, then publish the result as a project.
Signed-in event radar
Weekly event suggestions live in your builder dashboard and feed, where they can use saved stack, search, and opt-in source signals.
Builder proof loop
This loop lives inside the build workbench: compare the stack, save the path, hand context to Codex or Cursor, claim useful resources, and publish proof other builders can trust.
chooses the stack once
Compares backend, auth, vector, and data tradeoffs in one pass, then leaves with a Build Packet a coding agent can run immediately.
ships from Codex or Cursor
Keeps docs, env notes, eval checks, repo context, and launch proof together instead of scattering the handoff across six tools.
claims perks after intent
Searches, saves, claims credits, builds, and publishes without losing the proof trail that makes the project reusable after the event.
measures activation, not noise
Tracks first success, retained usage, source, and next action so developer growth compounds from product proof instead of a raw lead list.
The workbench should hold the full loop: choose the stack, save the decision, claim what helps, run the agent handoff, publish proof, and come back with the next build.
Product graph
0+
Indexed AI tools, alternatives, and backend paths ready for builder decisions.
Starter paths
0+
Build starts opened across RAG, auth, semantic search, agent workflows, and eval loops.
Activation metrics
0+
Save, packet, publish, perks, agent runs, shares, and return signals tracked in the builder loop.
Daily loops
0+
Acquisition, activation, proof, and follow-up checks already mapped into daily automation.
Choose stack
Compare the path before an agent writes code.
Claim resources
Match credits and programs to real build intent.
Publish proof
Turn shipped work into a public project page.
Bring signal back
Shares and return visits feed the builder graph.
Default Postgres, auth, storage, realtime, edge functions, and vector path.
InsForge
Agent-native backend path for builders who want CLI/MCP-driven auth, data, compute, and deploy decisions.
Neon
Postgres branching path for teams that need database isolation, previews, and safer agent changes.
Appwrite
Open-source backend suite for auth, database, storage, functions, and self-hosted control.
PocketBase
Fast local/single-binary backend for demos, internal tools, and lightweight shipped proof.
Convex
Reactive app backend for builders who want functions, data sync, and less glue code.
Source-linked
proof state