About ToolPassed
A directory of MCP servers, agent skills and AI agents, each listed with what we actually tested and when.
As of 2026-10-14: 37 MCP servers listed, 25 tested hands-on in at least one client, 15 passed every step in all 4 clients; 26 AI agents and 10 AI tools listed. See what changed.
Who runs it
ToolPassed is run by Scott Jones (@ATown_Jonesy on X), an independent publisher, and is built and operated with help from AI agents.
How testing works
We install each tool in real clients (Claude Code, Cursor, Claude Desktop and ChatGPT (Codex); Codex runs inside the ChatGPT desktop app), run a scripted task and publish the result with its date: passed, partial, failed or not tested. We never show a pass we didn't observe. We retest when a tool ships a new major version, when someone reports a problem, or when its link breaks. Our current picks are retested monthly. Every result shows the date it was tested. The details are on How we test.
Independence
Picks, ranks and results can't be bought. Sponsors and any featured spots are always labelled, and they never change a test result. See How we make money.
Can makers pay for a review?
Yes, for speed only. Free submissions are tested in the order received, first come, first served, with no promised timeline. A $99 priority review jumps that queue: your tool is hands-on tested within 7 days of payment or you get a full refund. It runs the same tests and gets the same honest verdict as every listing, which can be negative. It never buys a better score, placement or pick, and the listing is marked "Maker-submitted". No refund for an unfavorable verdict; full refund if we don't test it within 7 days.