Browser for your agent. Runs local in a window you can see
Watch your AI agent browse the web in real time. Pickle runs on your machine, using your real logins, with every action policy gated and logged as it happens. It compresses pages before your agent sees them, cutting token use by 32x. Works with Claude Desktop, Cursor, VS Code, and Codex CLI.
Hi Product Hunt! I’m Ismail, founder and CEO of Pickle.
I use agents for a lot of my own work, but whenever they use a browser they end up wasting tokens reading HTML, so I built a browser that simplifies webpage content for them. It's free, and compatible with Claude Desktop, Cursor, VS Code, and Codex CLI.
A few key things about it: Full visibility, always. Every action is policy gated, purchases and destructive steps need your approval, everything is logged, and you can watch it happen live on screen. Works with any agent. Claude Desktop, Cursor, VS Code, and Codex CLI today, and anything else that speaks MCP. Nothing to reconfigure. Works with weak models too. Pickle bundles Ollama, so a small local model can run the agent without an API key or the cloud. Weak models get constrained prompting that limits them to one choice at a time, so they don't hallucinate actions or element IDs. Stronger models get full multi-step planning. Reads pages efficiently. Pickle cuts token usage by 32x. Full browser, not just a wrapper. Tabs, history, bookmarks, real Chromium, plus persistent element memory across sessions and a shared notebook the agent and you both write to. Who is Pickle for? Anyone running an agent that needs to act in a browser and wants to see what it’s doing before it acts. If you connect Claude Desktop, Cursor, VS Code, or Codex CLI to real tasks, this is for you.
Would love to hear your feedback and questions!
Ismail
About Pickle Browser on Product Hunt
“Browser for your agent. Runs local in a window you can see”
Pickle Browser launched on Product Hunt on August 13th, 2026 and earned 85 upvotes and 1 comments, placing #26 on the daily leaderboard. Watch your AI agent browse the web in real time. Pickle runs on your machine, using your real logins, with every action policy gated and logged as it happens. It compresses pages before your agent sees them, cutting token use by 32x. Works with Claude Desktop, Cursor, VS Code, and Codex CLI.
On the analytics side, Pickle Browser competes within Developer Tools and Artificial Intelligence — topics that collectively have 993.7k followers on Product Hunt. The dashboard above tracks how Pickle Browser performed against the three products that launched closest to it on the same day.
Who hunted Pickle Browser?
Pickle Browser was hunted by Ismail Haddou. A “hunter” on Product Hunt is the community member who submits a product to the platform — uploading the images, the link, and tagging the makers behind it. Hunters typically write the first comment explaining why a product is worth attention, and their followers are notified the moment they post. Around 79% of featured launches on Product Hunt are self-hunted by their makers, but a well-known hunter still acts as a signal of quality to the rest of the community. See the full all-time top hunters leaderboard to discover who is shaping the Product Hunt ecosystem.
For a complete overview of Pickle Browser including community comment highlights and product details, visit the product overview.
Ismail Haddou
Hi Product Hunt! I’m Ismail, founder and CEO of Pickle.
I use agents for a lot of my own work, but whenever they use a browser they end up wasting tokens reading HTML, so I built a browser that simplifies webpage content for them. It's free, and compatible with Claude Desktop, Cursor, VS Code, and Codex CLI.
A few key things about it:
Full visibility, always. Every action is policy gated, purchases and destructive steps need your approval, everything is logged, and you can watch it happen live on screen.
Works with any agent. Claude Desktop, Cursor, VS Code, and Codex CLI today, and anything else that speaks MCP. Nothing to reconfigure.
Works with weak models too. Pickle bundles Ollama, so a small local model can run the agent without an API key or the cloud. Weak models get constrained prompting that limits them to one choice at a time, so they don't hallucinate actions or element IDs. Stronger models get full multi-step planning.
Reads pages efficiently. Pickle cuts token usage by 32x.
Full browser, not just a wrapper. Tabs, history, bookmarks, real Chromium, plus persistent element memory across sessions and a shared notebook the agent and you both write to.
Who is Pickle for?
Anyone running an agent that needs to act in a browser and wants to see what it’s doing before it acts. If you connect Claude Desktop, Cursor, VS Code, or Codex CLI to real tasks, this is for you.
Would love to hear your feedback and questions!
Ismail