This workflow covers bulk context extraction from your entire codebase. Your coding agent reads through your project and curates knowledge in multiple passes, automatically organizing everything into the context tree. For onboarding individual files (1-3 documents), see Onboard Existing Context.Documentation Index
Fetch the complete documentation index at: https://docs.byterover.dev/llms.txt
Use this file to discover all available pages before exploring further.
Prerequisites
- ByteRover CLI installed
- Project initialized (run
brvto start the REPL and initialize your project) - A codebase with existing patterns, conventions, or documentation worth capturing
Step 1: Curate from Your Codebase
- Via Coding Agent
- Commands
- Manual REPL
Copy this prompt and paste it into your coding agent’s chat (e.g., Cursor, Claude Code, etc.):Your coding agent will read through your codebase and run
brv curate multiple times to capture the knowledge. ByteRover automatically organizes everything into the context tree.Step 2: Review Changes
Verify what was added:- Commands
- Manual REPL
Step 3: Query to Verify
Test that your context is retrievable.- Via Coding Agent
- Commands
- Manual REPL
Copy this prompt and paste it into your coding agent’s chat:Your coding agent will execute
brv query with your query and return the relevant context.What Happened Behind the Scenes
- Autonomous Analysis: ByteRover analyzed your codebase content and detected relevant domains (code_style, testing, structure, api, etc.) using semantic understanding.
-
Intelligent Organization: Content was organized hierarchically into the context tree at
.brv/context-tree/, with appropriate domain and topic structure. -
Relation Mapping: ByteRover identified connections between related topics and created explicit
@domain/topicrelations to build a navigable knowledge graph. - Duplicate Prevention: Before creating new topics, ByteRover searched for existing related knowledge to prevent duplication and update existing topics when appropriate.
-
Facts Extraction: ByteRover extracted structured facts from your content — categorized as
personal,project,preference,convention,team,environment, orother— and stored them in the## Factssection of each knowledge file. See How Curation Works for details.
Customizing Your Bootstrap
You can customize how ByteRover processes your codebase: Focus on specific areas:Cloud Sync (Optional)
Push/pull requires a cloud account. See Local vs Cloud for setup.
- CLI
- TUI
Next Steps
- Onboard Existing Context - For individual files (1-3 docs)
- Query the Context Tree - Effective query patterns and tips
- Curate Context with Intelligence - Interactive vs autonomous modes, advanced options