slopmeter

17 agent-built projects, scored.

The tools the coding-agent ecosystem runs on — harnesses, skill packs, MCP servers, memory layers — each a real slopmeter github <owner/repo> run, scanned 2026-09-10 and committed as data. Alphabetical, not ranked. Run the same command on any of them and you get the same page.

repositorygradescorewhat costs the grade
affaan-m/ECCC71commit health
career-ops-hq/career-opsC73commit health
colbymchenry/codegraphD55commit health
DeusData/codebase-memory-mcpC69commit health
DietrichGebert/ponytailD54over-engineering
Egonex-AI/Understand-AnythingC65commit health
farion1231/cc-switchF48commit health
garrytan/gstackD51commit health
Graphify-Labs/graphifyD63over-engineering
headroomlabs-ai/headroomC67commit health
nexu-io/open-designC67commit health
NousResearch/hermes-agentC66commit health
obra/superpowersC65over-engineering
router-for-me/CLIProxyAPID62commit health
ruvnet/rufloD60commit health
stablyai/orcaD51commit health
thedotmack/claude-memD60commit health

A score is a measurement of a tree on a day, not a verdict on a project. The file-level findings are not published here — slopmeter prints those for the people who can act on them.

What slopmeter measures, and how