All projects
Code search for agents

Drawer.

Give your coding agent the context it needs.

Drawer searches your repository and returns source excerpts with file locations. Text search needs no model. Semantic search runs offline after model setup.

At a glance
Language
Rust
Focus
Code search for agents
Source
Codeberg

Rust · Local search · JSON output

Find code for your question
Text search works without a model. Semantic search needs model setup.

01 / The problem

Why it exists.

Coding agents can spend a lot of time opening files and collecting context. Drawer finds the relevant code locally and returns it within a size limit.

02 / The approach

How it works.

Ask a question or search for exact text. Drawer combines text search, syntax information, and optional local embeddings. Results include source excerpts, file locations, and hashes, with coverage notes and pagination.

Text search needs no model. Semantic search works offline after installing the model. Drawer also supports Git history and imported compiler references. Your coding agent handles reasoning, edits, and tests. Remote answers are optional and send selected evidence to your chosen provider.

code-searchcoding-agentslocal-firstcontext

03 / Contribute

Help improve Drawer.

Try Drawer, report a bug, or suggest a change in the repository.

Next project

Genesis/ Development environments