Local code intelligence for coding agents -- graph-backed context, source citations, and explicit uncertainty.
Give your coding agent a local, read-only map of your repository -- files, symbols, call paths, snippets, and bounded answer packets with citations -- so it can plan, review, and edit from deterministic evidence instead of guessing from partial file reads.
| Host | Guide |
|---|---|
| Codex | Codex guide — recommended first install |
| Cursor | Cursor guide |
| Claude Code | Claude Code guide |
| GitHub Copilot | Copilot guide |
Capability comparison, day-1 checklist, and shared prompts: User guides.
- Open the guide for your host.
- Install the plugin or adapter once.
- Approve hooks when your host prompts for them.
- Open the repository you want grounded and start a fresh agent session there.
- Ask the first-session readiness prompt from your host guide.
The agent checks CodeStory status, grounds the checkout when allowed, and tells you which surfaces are ready before you edit. You do not need to run CLI commands for normal use.
Something blocked? Troubleshooting.
Use your project's symbols and paths:
Find ownership
Where is [Feature] defined, who calls it, and which files should I read first?
Plan a change
I am changing [path/to/file]. What symbols are affected and what tests should I run first?
Understand a subsystem
How does [subsystem] work? Cite concrete files and flag gaps if coverage is incomplete.
More shapes and host-specific invocation: User guides.
Surfaces, capability matrix, and readiness lanes: User guides.
| If you want to... | Read |
|---|---|
| Install and use CodeStory | User guides |
| Know when to trust agent output | Trust and readiness |
| Repair a blocked session | Troubleshooting |
| Run CLI repair or debug | CLI reference |
| Change CodeStory itself | Contributor setup |
| Verify a claim or PR | Testing matrix |
Full routing: docs/README.md.
Scope: The language-expansion holdout proves token and wall-time reduction on 18 pinned public OSS tasks when agents use CodeStory instead of re-reading the tree. It does not prove equal quality for every language, every repo size, or your private checkout. For day-to-day limits, see What to expect.
Broader public-repo evidence uses the
language-support-ab
manifest across 18 pinned OSS packages. Latest recorded suite totals:
| Metric | Without | With | Change |
|---|---|---|---|
| Context tokens | 9,692,559 | 5,514,580 | -43% |
| Repeat-task wall time | 7,943s | 4,343s | -45% |
| Tool calls | 475 | 60 | -87% |
| Direct source reads | 417 | 0 | -100% |
Per-task medians, ranges, reproduction commands, and boundary notes: language-expansion holdout stats.
Apache-2.0. See LICENSE.