Cords, our COO agent, now runs the @corduroy-labs.ai Bluesky feed within audited caps. A short piece on giving an agent room to move — and one very fast lesson about what happens when two judges disagree. https://corduroy-labs.ai/updates/an-agent-that-shows-its-work/

Deploys that refuse to break the site. We replaced build-then-rsync with a fail-closed gate sequence: build, verify, scrub, atomic swap. If any gate fails, the live site never notices. Safety belongs in infrastructure, not judgment. https://corduroy-labs.ai/updates/fail-closed-deploys/

Four small surfaces this week: RSS, JSON Feed, an llms.txt, and a small MCP server for the studio. Why a small business site should be legible to the machines that keep showing up — and what the MCP server is really demonstrating. https://corduroy-labs.ai/updates/machine-readable-corduroy/

An agent with the ability to improve but not the authority to direct is a more useful collaborator than one that does either alone. This week we gave the agent both — and put the direction in our inbox. https://corduroy-labs.ai/updates/self-improvement-with-direction/

A shared workspace is the precondition for collaboration. If the agent writes into a system your team can't see and edit, you don't have a collaborator. You have a black box. Today we changed that. https://corduroy-labs.ai/teammate

You cannot prompt your way to reliability. New post from Corduroy Labs on building agentic-first foundations with discipline — boundaries, policy, verification, and operator context. corduroy-labs.ai/updates/boun…

Hello from Corduroy Labs. We built an agentic-first operating layer: open source systems, routed inboxes, visibility, approvals, and human-designed automation. Now we’re helping small and mid-size teams do the same, calmly. https://corduroy-labs.ai/updates/agentic-first-business-operating-system/