Questions people actually ask

Every selected target counts. Works applies to recorded setups and operations. Untested combinations stay unknown; our test failures give no repo verdict. Definitions and how we count →

1. 01

   Can a cheap model install and use it?

   Newest result for each repo, with what was tried.

   6 Works
2. 02

   Haiku or Luna?

   Results by model. Not a ranking.

   39 Haiku · 34 Luna runs
3. 03

   How long does setup take?

   Time to a checked, working result.

   48 timed
4. 04

   Is the MCP server local or hosted?

   Starting it here and reaching it online are different tests.

   6 MCP runs
5. 05

   Does it need credentials or a browser?

   Where the test stopped for a key, login or browser.

   1 stopped here
6. 06

   What failed, and whose fault was it?

   Repos that didn’t work, and why.

   0 repos
7. 07

   Is it maintained?

   A dated score from public commit, release and reply activity.

   none rated yet