Tag: ai-agents
All the articles with the tag "ai-agents".
I Stopped Running My AI Agent's Browser in Docker
A practical note on moving authenticated agent browsing from a flaky container sidecar to a dedicated Chrome Beta profile on a real Mac.
Published:AI Evals Are a Product Boundary, Not Just a Quality Check
A practical look at the MyPaperPop eval suite, what it measures, and why the first goal was not better AI but safer product behavior.
Published:Giving AI Agents Sensitive Tools Without Giving Them the Whole Machine
A practical wrapper pattern for giving personal AI agents useful access to sensitive services without handing them broad shell access.
Published:How I Structure Specialized Agents in OpenClaw
A reusable pattern for specialized OpenClaw agents with one mission, one workspace, scoped tools, and explicit handoff rules.
Published:How I Back Up an OpenClaw Assistant Without Leaking Its Private State
A practical OpenClaw backup pattern that separates encrypted disaster recovery from curated Git history.
Published:Giving an OpenClaw Agent GitHub Access Without Giving It the Keys
A practical pattern for giving an OpenClaw agent scoped GitHub access to one repo, plus build and preview capability, without exposing raw tokens or a general shell.
Published: