Zephium is an open-source browser and work environment built for humans and agents on the same surface. Its documentation describes a shared Tasks surface for both, an agentic browsing execution layer behind a feature called Work, a bounded browser session driven by a semantic locate and act controller, and work runtime subsystems covering execution, scoped runtime, composition, native lifetimes, persistence and retained resources. The project is explicit that code and tests are the evidence of present behavior, and each document owns a narrow claim: the security model controls guarantees, architecture controls the browser foundation, and the product system controls intent.
Be the first to rate Zephium. Your feedback helps other users decide.
People and agents work the same Tasks view instead of the agent driving a separate window.
Run agent browsing as a scoped, semantic locate and act session rather than an open-ended puppeteer script.
Read the security model to see which capabilities are enforced and which are gated off.
Similar tools in the same category — compare side-by-side
Jukebox OS is a deterministic flight recorder, ACID tool sandbox and time-travel runtime for multi-a...
Junior is an open-source AI junior engineer that runs on your Mac. It watches GitHub, Slack, Linear ...
AgentTrust is a verifiable zero-trust control plane, cryptographic black box and policy proxy that s...
Guides and comparisons about code tools
Eight launches that give agents a place to work: a browser, a terminal grid, a recording runtime, a process CLI and a reverse-engineering IDE.
Nine launches that keep work on your own hardware: a model storage server, offline file search, native dictation, ARM64 VMs and a single-binary coding agent.
Nine Show HN launches beneath the model: filesystems, snapshots, gateways and deploy tools that decide what an agent sees and what survives.
Nine launches that check agent work: per-call access policy, MCP server vetting, runtime verification, readable diffs and a hard stop at 95%.
Nine Show HN launches sitting under the agent: a model gateway, vectorless retrieval, on-device inference, an MCP browser layer and connectors for real systems.