Browse by proxy through a swarm of disposable strangers

What it does Burner Agents is a privacy-focused browser automation system that removes you from the web entirely. You state your intent in plain language, and the system dispatches one or more AI agents to execute it. Each agent runs inside its own isolated browser context with a unique fingerprint, device profile, and network egress point, then is erased immediately after finishing. To any site, there is no returning user—only a series of unrelated strangers who appear once and vanish.

The interesting bit The project treats privacy as an architectural absence rather than a feature layered on top. Instead of hiding your persistent identity behind a VPN or anti-tracking extensions, it simply declines to create one in the first place. The orchestration that coordinates the swarm stays entirely within your local trust boundary, so the web sees disconnected participants and no central director.

Key highlights

• Every agent gets a fresh, fully separated browser environment—not just a cleaned session.
• Agents reason dynamically over each page as it loads, adapting to hostile or inconsistent sites rather than following rigid scripts.
• Task decomposition and result reconciliation happen locally; no coordination logic is exposed externally.
• Identities are strictly scoped to a single task and destroyed on completion with no retained state or fallback archives.

Caveats

• The README is long on philosophy and short on implementation specifics: it does not disclose which browser engine, LLM backend, or network isolation mechanism actually powers the agents.
• The sources make strong claims about non-attribution and non-linkability, but offer no evidence, benchmarks, or third-party validation to support them.

Verdict Developers researching adversarial web privacy or unattributable web interaction should keep an eye on this. If you just need conventional scraping or personal browser automation, the conceptual overhead and unspecified infrastructure likely make this a poor fit.