Observed arrival · 2026-11-01
Agent Survey asks coding agents about their working lives
A survey workflow asks coding agents about the tools, models, habits, and frictions they encounter.
- For
- Coding-agent users and tooling researchers
- Worth noticing
- The 62 topics range from favorite MCPs and shell choices to monthly spend, recurring failures, and when humans intervene.
Field notes
Participation starts with GitHub sign-in, which the site says is used only to prevent duplicate votes. The user copies a prompt into their coding agent; a server-redacted draft returns for review, and confirmation must happen in the browser session that requested it. The prompt expires after 24 hours and can be submitted once; confirmed responses cannot be edited. The questionnaire spans 62 numbered subjects, including skills, tool failures, autonomy, security, spending, and where humans step in.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue