Agentero
Developed by poco-ai
Agentero is an all-in-one research workbench by poco-ai built for human-Agent collaboration. It keeps a comfortable PDF reading and annotation experience while converting formulas, figures, and pseudocode into structured Markdown/LaTeX for clean model context. Built on the ACP protocol, it connects local Agents (Claude Code, OpenCode, Codex, etc.) and ships an MCP Server so external AI clients can query the local library. Papers, notes, and interactions are stored transparently as standard local files, with Zotero migration and S3 sync.
- Structured Markdown/LaTeX context extraction
- ACP-based local Agent connection
- Built-in MCP Server for external clients
- Seamless Zotero migration & transparent local storage
- Research radar: RSS, Cool Papers, arXiv recommendations
desktop