Observed arrival · 2026-11-17
Ready for Agents tests which docs coding agents actually read
The early-access service runs Claude Code, Codex and Gemini on library tasks, testing documentation channels such as llms.txt, skills, docs MCP and AGENTS.md.
- For
- Library teams choosing documentation channels for coding agents
- Worth noticing
- In one reported library, an AGENTS.md brief raised current-client selection from 4/9 to 9/9; other gaps were labeled noise.
Field notes
Each run starts from a fresh project with no memory, turns on one documentation channel, and is repeated three times per agent; the stated design spans five channels and three agent products. The sample includes transcripts, hidden-test outcomes, per-run costs, and a significance note (p = 0.03 for one AGENTS.md comparison). It also records a practical failure mode: Codex listed a hosted docs MCP’s tools but did not call them in nine runs.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue