Hi, I'd like to suggest a complementary direction for antfarm.
Project:
https://github.com/AMAP-ML/SkillClaw
If antfarm is about spinning up an agent team quickly, I think a natural adjacent problem is what happens after those agents have been used for a while: their skills drift, duplicate, and fragment across contexts.
SkillClaw focuses on that post-task lifecycle layer by deduplicating, merging, improving, and sharing skills across agents/devices/teams.
So the fit here is: agent teams on the front end, long-term skill governance on the back end.
Paper:
https://arxiv.org/abs/2604.08377
Hi, I'd like to suggest a complementary direction for
antfarm.Project:
https://github.com/AMAP-ML/SkillClaw
If
antfarmis about spinning up an agent team quickly, I think a natural adjacent problem is what happens after those agents have been used for a while: their skills drift, duplicate, and fragment across contexts.SkillClaw focuses on that post-task lifecycle layer by deduplicating, merging, improving, and sharing skills across agents/devices/teams.
So the fit here is: agent teams on the front end, long-term skill governance on the back end.
Paper:
https://arxiv.org/abs/2604.08377