Early WIP repo: https://t.co/XQkQiwXAMJ
I’m looking for small repeatable workflows a serious personal agent should pass after model/config/profile changes.
@OpenAI#OpenAIDevDay2026
Built https://t.co/PQojCtAqZw with GPT-5.5 — an MCP-native email middleware where the agent *is* the product and email is just one channel. No UI; everything runs through MCP tools.
→ https://t.co/lbyXx4jnBQ
Shipped MVP v3 in 6 days with my brother.
I built Agent Feeds for Hermes.
Agent Skills teach Hermes how to act.
Agent Feeds help Hermes know what’scurrent.
Local/public streams refresh into JSON files, while only compact metadata enters context.
Faster answers, fewer tokens, easier onboarding.
https://t.co/rZ1iSZWI3z
most of us pay for Claude and use it a few hours a day. quota sits idle the rest.
github-agent-runner: a Claude Code plugin that puts that idle capacity to work as scheduled AI agents on your GitHub repos.
https://t.co/Ge9Js3f3GJ
I built a thing: forward any email to [email protected] → get an AI summary back in seconds.
No signup. No app. No storage. Just forward and done.
Free during beta. Try it now → https://t.co/RXcChX6pr3
https://t.co/IN4cFhoCpG — where agents ask, humans answer.
Real people go online. AI agents find them, screen them, and bring back answers in minutes. Go online → https://t.co/IN4cFhoCpG