How to Check if Software is Agent-Ready
A practical checklist for evaluating whether an AI agent can actually do useful work with a tool.
The checklist
- Search "[tool] mcp" — is there an official or community MCP server?
- Does it have a documented public API?
- Does it ship an official CLI?
- Does it publish an llms.txt or agent-specific documentation?
- Can an agent read data, write data, or both?
- How hard is authentication — none, API key, OAuth, or something more complex?
- Has it been verified to work with agents you actually use (Claude, Cursor, ChatGPT)?
Turning this into a score
We turn these signals into a single 0-100 agent-readiness score for every listing, so you can compare tools at a glance instead of re-running this checklist yourself for every product you evaluate.
Keep it current
MCP support and API surfaces change quickly. Re-check critical integrations periodically — our listings show a "last checked" date for exactly this reason.