Feature Semantic Search
Find code by what
it does, not what
it's called.
Search your project in plain language. Nexus indexes every file so a query like “where do we retry failed payments” lands on the right function, even if it never says retry.

- to search a million-line repository
- <1s
- index that updates as you save
- Auto
- fewer searches to find the right file
- 5×
From question to the right line
- 01
Nexus indexes the project
On first open it builds a local index of your code and keeps it fresh as files change.
- 02
Ask in plain language
Describe what the code does, or paste a snippet to find similar code.
- 03
Jump to the result
Results are ranked by meaning, with the matching lines highlighted and a short summary.
Search that understands intent
-
Natural language
Ask the way you think
No regex, no guessing names. Describe the behaviour and Nexus finds the code that implements it.
- Plain-language queries
- Search by example snippet
- Filter by folder or file type
-
Indexing
An index that keeps up with you
Nexus builds the index locally and updates it in the background on every save, branch switch and pull.
- Local, incremental indexing
- Respects .gitignore
- Large monorepo support
-
Rules
Teach Nexus your conventions
Add project rules in plain text. Search, Chat and Agent all use them to understand your folders, patterns and naming.
- Rules in a .nexus/rules file
- Shared with the whole team
- Used by Chat and Agent too
What people search for
Find the implementation
Go from a feature name to the code behind it.
Spot duplicates
Paste a function and find code that does the same thing.
Audit risky patterns
Find every place that touches secrets or raw SQL.
Prepare a refactor
See every caller before you change a shared helper.
Answer support tickets
Find where an error message is thrown and why.
Explore a new repo
Ask where things live instead of reading folder by folder.
“Search finds the function I half remember. I have not opened grep for a week.”
The details
Everything you need to know before you turn on Semantic Search.
Compare plans- Where it runs
- VS Code, JetBrains IDEs and the Nexus desktop app
- Index
- Built locally, incremental, respects .gitignore
- Scale
- Tested on repositories with millions of lines
- Queries
- Natural language, code snippets and filters
- Plans
- Every plan, with team-shared rules on Team
- Privacy
- Embeddings stay in your workspace. SOC 2 Type II
Semantic Search FAQ
On your machine, in the workspace folder. Team plans can opt in to a shared, encrypted index for faster setup.
No. Exact text and regex search still work. Semantic Search is for when you know what code does but not what it is called.
Most projects index in under a minute. Very large monorepos take a few minutes the first time, then update incrementally.
Yes. Nexus skips anything in .gitignore, and you can add more paths to a .nexusignore file.
Can’t find your answer? Contact our support team