11 labs · 2 X roundups a day
AgentLensHQ tracks 8,500+ AI projects day by day and catches the ones breaking out before they trend. It reads Hacker News, the researchers on X (formerly Twitter), and the labs themselves — OpenAI, Anthropic, DeepMind and the rest. Everything that matters is filed as a readable dispatch, and served to your agents over MCP.
Hacker News · X · Labs
OpenAI · Anthropic · Google DeepMind · xAI · DeepSeek · Qwen · Mistral AI · Hugging Face · vLLM · Ollama · BAIR
LLM / Agents · Image · Video · Audio / Speech · Vision · Multimodal · Robotics · Infra / Tooling · Other
↓ Projects
A ranking every week, plus teardowns that go down to the implementation.
AI Open-Source RadarThe Top Slot Has Changed Hands Every Week; the Rest of the List Has NotEvery issue so far has had a different leader, and this week's had never appeared on the board at all — yet eight of the ten names carried straight over. Beneath the churn, Ponytail's climb and DeepSeek Harness's slide have ended in the same place.AI Agent Harnesses, Read From SourceWhat a Harness Actually Is, in 190 Lines of CodeThe word harness is everywhere now, and almost nobody points out that it names two things at once: the program that runs the loop, and the files you prepare for it inside your own repository. Keep those fused and every harness article you read will, at some point, stop lining up. This one pulls them apart using the main loop of mini-swe-agent — 190 lines, short enough to read in one sitting, and each of the five subsystems (instructions, tools, environment, state, feedback) lands on a line number you can point at.The community has already voted. We read the comments too — a story whose discussion we could not fetch never becomes a dispatch at all. And it is not written once and left: as the discussion keeps heating up, the dispatch is rewritten with whatever the thread has since said.
No more scrolling the timeline, and no more being kept inside the algorithm's bubble. We filter it for you every day and file two roundups of what was actually worth reading.
No more opening a dozen official blogs every morning. First-hand releases from OpenAI, Anthropic, DeepMind and the rest, each with its substance pulled out.
Every source is structured and vectorised, then exposed over the Model Context Protocol — so your agents query fresh, grounded context instead of stale training data.
# add the AgentLensHQ MCP server claude mcp add --transport http agentlens \ https://api.agentlenshq.com/mcp
