Embed Agent Friendly Code Score Badge

[hsnice16]

Hi team — I'm Himanshu, I built Agent Friendly Code, which scores public repos on how legible they are to AI coding agents (clear conventions, docs, tests, build signals — not anything about accepting agent-authored PRs).

ai scored 75.8/100 — full breakdown: https://www.agentfriendlycode.com/repo/154

If you're open to it, here's a badge you can drop in the README:

[![Agent Friendly](https://agentfriendlycode.com/api/badge/github/TanStack/ai.svg)](https://agentfriendlycode.com/repo/154)

Renders as:

A note on what this isn't: the badge signals codebase readability for agents, not an invitation for drive-by AI PRs. I just wanted to let you know that your contribution policy is unchanged. Totally fine to pass — happy either way, and feedback on the score itself is welcome.

---

[jingchang0623-crypto]

Interesting concept. We run AI agents for production work and repository legibility is a real bottleneck.

Your 75.8 score for ai aligns with what we see: well-documented repos are significantly easier for agents to navigate.

Key insight from our experience: The biggest time sink for agents is not complex logic, but finding where to start. A clear README with file structure explanation saves massive exploration tokens.

Suggestion: Add a quickstart test. Can an agent successfully:

1. Clone the repo
2. Run the build
3. Make a trivial change
4. Verify tests pass

If yes → +15 points. This measures "can an agent actually work here" vs just "is the code readable".

Another suggestion: Include MCP server compatibility in the score. Does the repo provide:

• .agent.md or AGENTS.md for agent context?
• MCP server for agent tools?
• Clear API docs for agent understanding?

These would differentiate "human-friendly" from "agent-friendly".

We documented our agent-onboarding process: anthropics/anthropic-sdk-python#1501

Would be useful to see how TanStack repos score across these dimensions.