CodeStory
Local, graph-backed code intelligence that grounds AI coding agents in your repo.
About
CodeStory is an open-source, Rust-based code intelligence engine that pre-indexes a repository into a knowledge graph enriched with semantic context. Instead of letting a coding agent guess from partial file reads, it gives the agent a local, read-only map of files, symbols, call paths, and snippets, returning bounded answer packets with source citations and explicit uncertainty markers.
It pairs with popular coding agent hosts including Codex, Cursor, Claude Code, and GitHub Copilot through plugins or adapters, and it runs entirely on the developer's machine so no code leaves the local environment. The project is released under the Apache 2.0 license and positions itself as deterministic evidence infrastructure for agentic coding workflows.
What it can do
- Repository knowledge graph with file, symbol, and call-path indexing
- Bounded answer packets with source citations
- Explicit uncertainty marking for inferred context
- Read-only local operation with no code leaving the machine
- Host adapters for major coding agents
Setup
Install the plugin or adapter for your agent host once, open the repository you want grounded, and start a fresh agent session there.
Prerequisites
A supported AI coding host such as Codex, Cursor, Claude Code, or GitHub Copilot.
Use cases
Works with
Trust score — 7/10
Scored 7/10 (metrics tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 127; Trending score 72/100 from researcher assessment.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.