Writing
Field notes on design systems.
Features explain real systems. Workflows show the steps. Field Notes report what source checks and implementation exercises reveal, including their limits.
Post types
03- 01
Features
Deep dives on specific design systems and the decisions behind them. One system, one angle, end-to-end.
- 02
Workflows
Practical, copy-the-steps guides — token pipelines, audits, governance setup, AI-agent integration.
- 03
Field Notes
Source checks and findings from real implementations, with exact versions, methods and limits.
Recent
03- Field NotesSep 10, 20267 min read
What changed when we read the repos instead of the docs
The September re-audit of the Agent-Ready Index read each system's repository tree rather than its documentation. Unknown cells fell from 82 to 48, Primer's MCP server was corrected from no to yes, Carbon took the top score, and three of the auditor's own positives turned out to be wrong.
- Field NotesSep 5, 20265 min read
Three style references, checked against their source
An 80-field review of Carbon, Mantine and a legacy shadcn zinc snapshot found mismatched values, lost scale information and missing context. Exact sources and corrections are included.
- Field NotesJul 6, 20267 min read
What we learned auditing 37 design systems for AI-readiness
Five signals, 37 systems, first-party evidence only. The patterns that emerged: llms.txt is the easy race, first-party MCP tracks commercial pressure, the shadcn registry is a category of one, and DTCG conformance is nearly invisible.
Newsletter
The Agent-Ready Brief
Quarterly Agent-Ready Index changes and occasional field notes. No filler, unsubscribe anytime.
The email list isn't open yet — RSS carries every post in the meantime.