Launching today

baoer_signal_grep
A tiny local librarian for AI agents
1 follower
A tiny local librarian for AI agents
1 follower
Give AI agents a clearer way to search local code, docs, notes, and logs. Get focused passages or compact file maps, stable continuation cursors, and explicit coverage limits. Hybrid search separates exact matches from local semantic candidates. Open source, with source navigation for Pi, OMP, and MCP clients such as Codex. GitHub: https://github.com/xcjy8bao/baoer_signal_grep npm: https://www.npmjs.com/package/baoer_signal_grep

Hi Product Hunt, I’m baoer, the maker of baoer_signal_grep.
I built it after watching capable coding agents lose their place while searching a repository: open a file, dump a pile of matches, start over, then sound certain about something they hadn’t finished checking. I wanted to give them a small local librarian with a bookmark.
Small searches return the passage and its location. Broader searches can start with a compact file map. Follow-up requests continue through the retained snapshot, while limits and incomplete coverage stay visible. Hybrid search puts exact evidence first and labels local semantic candidates as leads to inspect.
It works with code, docs, notes, and logs, and includes source navigation for deeper investigations. You can use it through Pi, OMP, or an MCP client such as Codex. The project is open source and available on npm.
Try it or inspect the source:
GitHub: https://github.com/xcjy8bao/baoer_signal_grep
npm: https://www.npmjs.com/package/baoer_signal_grep
One boundary matters: a saved cursor preserves an investigation’s snapshot; it isn’t a live view of later file changes. Agents still need to read the evidence and judge what it supports.
I’d love to hear where your agent loses the thread: long logs, a changing worktree, an awkward search, or a result that looks more conclusive than it should. A small reproduction would be especially helpful.
The gallery artwork is an AI-generated concept illustration, not a product screenshot.
Thanks for taking a look. I’m here to answer questions and learn from the rough edges.