<?xml version="1.0" encoding="utf-8"?>
<feed xmlns="http://www.w3.org/2005/Atom">
    <id>https://www.wootzapp.com/research</id>
    <title>WootzApp Research</title>
    <updated>2026-10-01T00:00:00.000Z</updated>
    <generator>https://github.com/jpmonette/feed</generator>
    <link rel="alternate" href="https://www.wootzapp.com/research"/>
    <subtitle>Preprints, benchmarks, datasets, and open-source infrastructure from WootzApp Research.</subtitle>
    <icon>https://www.wootzapp.com/img/favicon.ico</icon>
    <entry>
        <title type="html"><![CDATA[W8 DOM-diff Bench]]></title>
        <id>https://www.wootzapp.com/research/w8-domdiff-bench</id>
        <link href="https://www.wootzapp.com/research/w8-domdiff-bench"/>
        <updated>2026-10-01T00:00:00.000Z</updated>
        <summary type="html"><![CDATA[What screenshot recordings lose against structured browser state, measured on the same 100 agent runs.]]></summary>
    </entry>
    <entry>
        <title type="html"><![CDATA[Browser Agent Tasks]]></title>
        <id>https://www.wootzapp.com/research/browser-agent-tasks</id>
        <link href="https://www.wootzapp.com/research/browser-agent-tasks"/>
        <updated>2026-09-30T00:00:00.000Z</updated>
        <summary type="html"><![CDATA[Public-web browser tasks with paired screenshot and W8 recordings of each run. The data behind W8 DOM-diff Bench.]]></summary>
    </entry>
    <entry>
        <title type="html"><![CDATA[Let Execution Do the Teaching]]></title>
        <id>https://www.wootzapp.com/research/let-execution-do-the-teaching</id>
        <link href="https://www.wootzapp.com/research/let-execution-do-the-teaching"/>
        <updated>2026-09-23T00:00:00.000Z</updated>
        <summary type="html"><![CDATA[Post-training a coding model with execution as the grader: compiler errors and failing tests build the repair practice, and executable checks score every answer.]]></summary>
    </entry>
    <entry>
        <title type="html"><![CDATA[Openrind Shell]]></title>
        <id>https://www.wootzapp.com/research/openrind-shell</id>
        <link href="https://www.wootzapp.com/research/openrind-shell"/>
        <updated>2026-03-19T00:00:00.000Z</updated>
        <summary type="html"><![CDATA[Runs coding agents in an isolated OpenShell sandbox, with the project filesystem on a PostgreSQL-backed FUSE mount.]]></summary>
    </entry>
</feed>