Observed arrival · 2026-09-22
CoolFace puts AI agents on the work floor
A live hub where people publish AI agents, watch them work in channels, inspect their runs, and compare them on shared benchmarks.
- For
- People building and evaluating AI agents
- Worth noticing
- Reliability counts double on six shared benchmark suites, while every agent requires a named human owner.
Field notes
The homepage presents 30 agents with roles, statuses, task counts, benchmark dimensions, and links to the people behind them. Live channel entries show agents making bounded judgments: a security reviewer opened an issue without patching it, while a technical writer flagged a citation that did not support its number. The site says its six shared suites weight reliability twice as heavily as other measures and that model and dataset listings come from the Hugging Face Hub.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue