Byterover Cli
ByteRover CLI (brv) - The portable memory layer for autonomous coding agents (formerly Cipher)
Last verified:
What is Byterover Cli?
ByteRover CLI is a local-first AI context engineering tool designed for coding agents. It gives your AI coding agents persistent memory stored locally on your machine, allowing them to remember patterns, decisions, and learnings across sessions. The tool curates context from your codebase automatically through agent pairing and enables team members to sync shared context so everyone works with the same knowledge.[web:1][web:5]
Key features include a Context Tree structure for organized memory management stored as human-readable markdown files, support for 18+ external LLM providers (OpenRouter, Anthropic, OpenAI, and more) with bring-your-own-key flexibility, and Git-Semantic version control for branching, merging, and syncing context trees. The CLI works seamlessly with 15-20+ coding platforms including Claude Code, Cursor, Codex, Windsurf, Cline, Gemini CLI, and GitHub Copilot.[web:1][web:2][web:5]
ByteRover CLI is designed for developers and AI coding agents who need persistent long-term memory for agentic coding. It serves individual power users working on complex projects requiring context continuity, as well as collaborative teams that need shared knowledge bases. The tool is especially valuable for teams using multiple AI coding tools who want to switch between tools without losing context.[web:2][web:5]
The daemon-first architecture enables instant startup with background execution, while the BRV Hub provides a community marketplace for pre-built agent skills and context bundles. Visual mode with Tree View and Grid View plus full-text search with filtering makes browsing and organizing context data intuitive.[web:7][web:10]
Byterover Cli pricing
Pricing model: Freemium
ByteRover offers three pricing tiers: **Free Plan** - Local-first, forever free: - Limited credits with ByteRover provider - Markdown-backed context store - 100% local - no cloud, no accounts required - Bring Your Own Key (use your own LLM API keys) - Zero telemetry - Community support - Full CLI access **Pro Plan** - $19/month (or $15/month billed annually, saving 21%): - Everything in Free, plus: - 4,500 credits per month - 1 user - no team invites allowed - 500 MB storage - Cloud Sync - Context Management feature - 48-hour Email SLA - Premium community access **Team Plan** - $35 per user/month (or $28/month per user billed annually): - Everything in Free, plus: - 4,500 credits per user per month - 5 GiB storage - Cloud Sync - Context Management - 24-hour Email SLA - Priority agent access - Premium community access - Popular for collaborative teams **Enterprise Plan** - Custom pricing: - Custom model allocation - Custom connectors - SSO/SAML and SOC 2 compliance - RBAC (Role-Based Access Control) - Data Residency options - On-prem gateway - Dedicated CSM and Slack channel - Audit logs
Byterover Cli pros
- Local-first with no cloud account required for basic usage
- Persistent memory across AI coding agent sessions
- Works with 15-20+ coding platforms including Claude Code, Cursor, Windsurf
- Bring Your Own Key (BYOK) supports 18 external LLM providers
- Context Tree stored as human-readable, git-friendly markdown files
- Git-Semantic version control for branching, merging, and team sync
- No telemetry - complete privacy for local usage
- Daemon-first architecture for instant startup and auto-start on demand
- BRV Hub community marketplace for agent skills and context bundles
- Visual mode with Tree View and Grid View for context management
- Full-text search with filtering capabilities for context retrieval
- Zero setup required - installs via single curl script or npm
- Agentic Search replaces traditional Vector DB for better retrieval precision
- 92.2% retrieval accuracy on LoCoMo and LongMemEval benchmarks
- Automatic memory flush before token limit fills to prevent data loss
- Cross-platform support: macOS, Windows, Linux, and WSL
- Free tier is forever free with full CLI access locally
Byterover Cli cons
- WSL has occasional performance and stability issues compared to native
- WSL requires coding agents to access the WSL environment specifically
- ByteRover built-in LLM requires account authentication and has limited free credits
- Cloud sync features require creating a ByteRover account
- Team invites not available on Pro plan - only 1 user
- Pro plan storage limited to 500 MB
- External LLM provider API keys required (not included in pricing)
- Migration from V2 MCP-based to V3 CLI-based requires configuration changes
- Codex users need network access approval for ByteRover commands
- Context Management feature only available on Pro and Team plans
- Yearly billing required for 21% discount on paid plans
- Enterprise features like SSO and SOC 2 require custom pricing
- Learning curve for Git-Semantic version control concepts
- Local credential storage requires libsecret-1-dev on Linux distributions
Frequently asked questions about Byterover Cli
Do I need a ByteRover account to use the CLI?
No, a ByteRover account is not required for local usage. The CLI works 100% locally with no cloud or accounts needed. You can connect any of the 18 external LLM providers using just your API key. A ByteRover account is only required if you want to use the built-in LLM (which has limited free credits) or access cloud sync features for team collaboration.
What AI coding tools does ByteRover work with?
ByteRover works seamlessly with 15-20+ coding platforms including Claude Code, Cursor, Windsurf, Codex, Gemini CLI, Goose, GitHub Copilot, Cline, Claude Desktop, Kilo Code, Roo Code, Amp, Kiro, Junie, Antigravity, Qoder, Trae.ai, Auggie CLI, OpenClaude, OpenClaw, OpenCode, and Warp. Different connector types are automatically selected based on the agent - Skill for most agents, MCP for 5 agents like Augment Code and Claude Desktop.
How does the Context Tree work?
The Context Tree is your project's knowledge base stored as markdown files in the .brv/context-tree/ directory. It is human-readable, git-friendly, and hierarchically organized by domain and topic (e.g., testing, architecture, database). ByteRover automatically consolidates, synthesizes, and prunes the context tree through its Dreaming feature, and uses Agentic Search for task-aware, question-specific retrieval instead of traditional Vector DBs.
Can I use ByteRover completely offline?
Yes, everything works offline by default. The CLI is local-first with no cloud required. You can use external LLM providers with your own API keys, and all context is stored locally. Cloud sync for team collaboration is entirely optional and only needed when you want to share context with teammates or access it from multiple machines.
What is Git-Semantic version control?
Git-Semantic version control is ByteRover's version control system for your context tree. It allows you to branch, merge, and sync context with teammates using familiar Git-like commands. You can initialize version control with brv vc init, stage changes with brv vc add, commit with brv vc commit, and push to cloud for team collaboration. It tracks changes to your context tree with semantic versioning.
How do I connect an external LLM provider?
You can connect external LLM providers via CLI or TUI. Using CLI: brv providers connect openrouter --api-key your-key --model anthropic/claude-sonnet-4.5, or brv providers connect openai --oauth for OAuth authentication. Using TUI: run /providers to select a provider, then /model to choose a model. Supported providers include OpenRouter, Anthropic, OpenAI, and 15 others. No ByteRover account needed - just your API key.
What is the difference between local and cloud usage?
Local usage is 100% offline with no cloud or account required - all context stays on your machine. Cloud adds team sync via Git-Semantic version control, multi-machine access, and backup without changing your local workflow. Cloud features include Cloud Sync, Context Management visual interface, and team collaboration features. The Free plan is entirely local, while Pro and Team plans add cloud capabilities.
How does context curation work?
Context curation happens automatically when your coding agent analyzes the codebase and calls brv curate, or manually via /curate in TUI or brv curate "description" in CLI. ByteRover organizes curated context into structured domains inside the local context tree. The curation engine automatically extracts and stores knowledge, while Session Learning extracts durable knowledge from sessions and Dreaming consolidates and prunes the tree.
How do I query context from my agents?
Your coding agent can query context by running brv query "your question" automatically when prompted. You can also query manually via /query in TUI or brv query "What are the unit testing strategies in this project?" in CLI. ByteRover retrieves the most relevant curated context using its query engine with Agentic Search, returning precise knowledge without irrelevant memory chunks.
What happens if I'm migrating from ByteRover V2?
ByteRover V3 introduces a new CLI-based architecture replacing the V2 MCP server approach. You need to remove V2 MCP configurations from your coding tools - agent-specific migration guides are provided. The legacy V2 web application is still accessible at https://app-v2.byterover.dev/ or from the new V3 web app dropdown by selecting Access ByteRover 2.0 (Legacy). A single curation command can convert existing flat memory logs into an organized Context Tree.