freshcrate
Skin:/
Home > MCP Servers > byterover-cli

byterover-cli

ByteRover CLI (brv) - The portable memory layer for autonomous coding agents (formerly Cipher)

Why this rank:Strong adoptionRecent releaseHealthy release cadence

Description

ByteRover CLI (brv) - The portable memory layer for autonomous coding agents (formerly Cipher)

README

ByteRover CLI

ByteRover Logo

Interactive REPL CLI for AI-powered context memory

License Version Downloads Documentation Discord

Overview

ByteRover CLI (brv) gives AI coding agents persistent, structured memory. It lets developers curate project knowledge into a context tree, sync it to the cloud, and share it across tools and teammates.

Run brv in any project directory to start an interactive REPL powered by your choice of LLM. The agent understands your codebase through an agentic map, can read and write files, execute code, and store knowledge for future sessions.

๐Ÿ“„ Read the paper for the full technical details.

Or download our self-hosted PDF version of the paper here.

Key Features:

  • ๐Ÿ–ฅ๏ธ Interactive TUI with REPL interface (React/Ink)
  • ๐Ÿง  Context tree and knowledge storage management
  • ๐Ÿ”€ Git-like version control for the context tree (branch, commit, merge, push/pull)
  • ๐Ÿค– 18 LLM providers (Anthropic, OpenAI, Google, Groq, Mistral, xAI, and more)
  • ๐Ÿ› ๏ธ 24 built-in agent tools (code exec, file ops, knowledge search, memory management)
  • ๐Ÿ”„ Cloud sync with push/pull
  • ๐Ÿ‘€ Review workflow for curate operations (approve/reject pending changes)
  • ๐Ÿ”Œ MCP (Model Context Protocol) integration
  • ๐Ÿ“ฆ Hub and connectors ecosystem for skills and bundles
  • ๐Ÿค Works with 22+ AI coding agents (Cursor, Claude Code, Windsurf, Cline, and more)
  • ๐Ÿข Enterprise proxy support

Benchmark Results

All benchmarks are run using the production byterover-cli codebase in this repository - no separate research prototype.

We evaluate on two long-term conversational memory benchmarks:

  • LoCoMo - ultra-long conversations (~20K tokens, 35 sessions) testing single-hop, multi-hop, temporal, and open-domain retrieval.
  • LongMemEval-S - large-scale benchmark (23,867 docs, ~48 sessions per question) testing 6 memory abilities including knowledge update, temporal reasoning, and multi-session synthesis.

LoCoMo - 96.1% overall accuracy (1,982 questions, 272 docs).

LongMemEval-S - 92.8% overall accuracy (500 questions, 23,867 docs).

LongMemEval-S Benchmark by Category

All metrics are LLM-as-Judge accuracy (%). See the paper, or self-hosted PDF for full details.

Quick Start

Shell Script (macOS & Linux)

No Node.js required - everything is bundled.

curl -fsSL https://byterover.dev/install.sh | sh

Supported platforms: macOS ARM64, macOS x64 (Intel), Linux x64, Linux ARM64.

npm (All Platforms)

Requires Node.js >= 20.

npm install -g byterover-cli

Verify

brv --version

First Run

cd your/project
brv

The REPL auto-configures on first run - no setup needed. Type / to discover all available commands:

/curate "Auth uses JWT with 24h expiry" @src/middleware/auth.ts
/query How is authentication implemented?

Web UI Development

The web UI supports a local-first development flow for the shared component library.

npm run dev:ui uses the git submodule at packages/byterover-packages/ui so edits to shared UI components hot-reload immediately in Vite.

# Clone with submodules, or initialize them after clone
git clone --recurse-submodules <repo-url>
# or
git submodule update --init --recursive

# Install dependencies
npm ci

# Start or restart the daemon
./bin/dev.js restart

# Start the web UI in local development mode
npm run dev:ui

Notes:

  • Edit shared components in packages/byterover-packages/ui/src.
  • npm run dev:ui uses the submodule source.
  • npm run build:ui uses the installed package path.
  • If /api/ui/config or transport bootstrap fails, restart the Vite dev server after restarting the daemon.

ByteRover Cloud

ByteRover Cloud is a hosted platform for teams to sync, share, and manage context knowledge across projects and machines. Everything works locally by default - Cloud adds collaboration and persistence without changing your workflow.

Try ByteRover CloudSign in with your ByteRover account via /login (TUI) or an API key (brv login) to get started.

  • ๐Ÿ”„ Team context sync โ€” push and pull shared knowledge across teammates
  • ๐Ÿ“‚ Shared spaces โ€” organize context across multiple projects and teams
  • ๐Ÿ’ป Multi-machine access โ€” sync your context tree across devices with cloud backup
  • ๐Ÿง  Built-in hosted LLM โ€” start immediately with limited free usage
  • ๐Ÿ‘ฅ Team management โ€” manage members, spaces, and permissions via the web app
  • ๐Ÿ“Š Usage analytics โ€” track seat allocation and monthly credit consumption
  • ๐Ÿ”’ SOC 2 Type II certified infrastructure with privacy mode

CLI Usage

Core Workflow

brv                  # Start interactive REPL
brv status           # Show project and daemon status
brv curate           # Add context to knowledge storage
brv curate view      # View curate history
brv query            # Query context tree and knowledge
brv review pending   # List pending review operations
brv review approve   # Approve curate operations
brv review reject    # Reject curate operations

Sync (Legacy)

brv push             # Legacy โ€” migrate or snapshot context to cloud
brv pull             # Legacy โ€” restore context from cloud snapshot

Use brv vc push / brv vc pull for version-controlled sync going forward.

Version Control

brv vc init              # Initialize version control for context tree
brv vc status            # Show version control status
brv vc add               # Stage files for the next commit
brv vc commit            # Save staged changes as a commit
brv vc log               # Show commit history
brv vc branch            # List, create, or delete branches
brv vc checkout          # Switch branches
brv vc merge             # Merge a branch into the current branch
brv vc clone             # Clone a ByteRover space repository
brv vc push              # Push commits to ByteRover cloud
brv vc pull              # Pull commits from ByteRover cloud
brv vc fetch             # Fetch refs from ByteRover cloud
brv vc remote            # Show current remote origin
brv vc remote add        # Add a named remote
brv vc remote set-url    # Update a remote URL
brv vc config            # Get or set commit author
brv vc reset             # Unstage files or undo commits

Providers & Models

brv providers list       # List available LLM providers
brv providers connect    # Connect to an LLM provider
brv providers switch     # Switch active provider
brv providers disconnect # Disconnect a provider
brv model list           # List available models
brv model switch         # Switch active model

Hub & Connectors

brv hub list             # List available hub packages
brv hub install          # Install a hub package
brv hub registry add     # Add a custom registry
brv hub registry list    # List registries
brv hub registry remove  # Remove a registry
brv connectors list      # List connectors
brv connectors install   # Install a connector

Spaces (Deprecated)

brv space list       # Deprecated โ€” use web dashboard
brv space switch     # Deprecated โ€” use brv vc clone

Other

brv mcp              # Start MCP server
brv login            # Authenticate to ByteRover
brv logout           # Disconnect and clear credentials
brv locations        # List registered projects
brv restart          # Restart daemon
brv debug            # Debug mode

Run brv --help for the full command reference.

Supported LLM Providers

ByteRover CLI supports 18 LLM providers out of the box. Use brv providers connect to set up a provider and brv providers switch to change the active one.

Provider Description
Anthropic Claude models
OpenAI GPT models
Google Gemini models
Groq Fast inference
Mistral Mistral models
xAI Grok models
Cerebras Fast inference
Cohere Command models
DeepInfra Open-source model hosting
OpenRouter Multi-provider gateway
Perplexity Search-augmented models
TogetherAI Open-source model hosting
Vercel AI SDK provider
Minimax Minimax models
Moonshot Kimi models
GLM GLM models
OpenAI-Compatible Any OpenAI-compatible API
ByteRover ByteRover's hosted models

Worktrees and Knowledge Sources

Vocabulary

  • Worktree link โ€” a subdirectory pointer to a parent project (brv worktree)
  • Source โ€” a read-only reference to another project's knowledge (brv source)
  • Origin โ€” where an indexed search result came from (local vs shared)

ByteRover can run from a linked subdirectory without creating a nested .brv/.

  • projectRoot: the directory that owns .brv/config.json
  • worktreeRoot: the linked worktree directory, or projectRoot when unlinked
  • clientCwd: the shell cwd where you ran brv

When you run brv query or brv curate from a linked worktree, implicit defaults use worktreeRoot so scope stays stable even if clientCwd drifts deeper into the package. Explicit relative paths that you pass yourself, such as brv curate -f ./src/auth.ts, still resolve from clientCwd to match normal shell behavior.

Use brv worktree add from the project root to register a subdirectory (or sibling) as a worktree. This creates a .brv pointer file in the target directory that redirects to the parent project โ€” the same pattern git uses for git worktree. Use brv worktree remove to unregister, and brv worktree list to inspect. To search another project's knowledge from here, use brv source add <path> (with brv source list / brv source remove to inspect or detach).

Documentation

Visit docs.byterover.dev for full guides on setup, integrations, and advanced usage.

Topic Description
Getting Started Installation, first run, and basic usage
Cloud Sync Push/pull workflows and team sharing
Version Control Context tree branching, commits, and collaboration
LLM Providers Provider setup and model configuration
AI Agent Integrations Using ByteRover with Cursor, Claude Code, Windsurf, etc.
Hub & Connectors Skills, bundles, and the connector ecosystem
CLI Reference Run brv --help

Contributing

We welcome contributions! See our Contributing Guide for development setup, coding standards, and the PR workflow.

Community & Support

ByteRover CLI is built and maintained by the ByteRover team.

  • Join our Discord to share projects, ask questions, or just say hi
  • Report issues on GitHub
  • If you enjoy ByteRover CLI, please give us a star on GitHub โ€” it helps a lot!
  • Follow @kevinnguyendn on X

Contributors

Contributors

Star History

Star History Chart

License

Elastic License 2.0. See LICENSE for full terms.

Release History

VersionChangesUrgencyDate
v3.16.1ByteRover CLI 3.16.1 is a small patch release with one fix: a valid login is no longer misread at startup, so a signed-in session stays signed in across restarts. No more being bounced to the provider picker or asked to run `brv login` again on every launch. Safe to upgrade, with no behavior change and nothing to reconfigure. ## What's Changed ### ๐Ÿ› Fixed * **Stay signed in across restarts.** A valid login could be misread when the CLI started up, which dropped you into the provider piHigh5/27/2026
v3.16.0ByteRover CLI 3.16.0 lets you cancel long-running tasks from anywhere and gives you a switch to turn off update prompts. It also fixes a setup error that blocked custom models on several third-party AI providers, and refreshes dependencies to clear security advisories. Safe to upgrade, with no behavior change unless you opt in. ## What's Changed ### โœจ Added * **Cancel a running task from anywhere.** `brv curate`, `brv query`, and `brv dream` now accept `--cancel <taskId>`, so you can stHigh5/25/2026
v3.15.0ByteRover CLI 3.15.0 introduces user-configurable settings via the new `brv settings` command, so you can tune things like agent pool size, concurrent tasks per project, and the LLM iteration budget without editing source or rebuilding. The `brv webui` Configuration page is reorganized into three sections (General, Connectors, Version Control) to make room for the new controls, folder metadata now renders in the Contexts detail panel, and the `--timeout` flag is deprecated in favor of `llm.iteraHigh5/20/2026
v3.13.0ByteRover CLI 3.13.0 reworks the billing experience: your remaining ByteRover credits now show in the REPL and web headers, and connecting a provider lets you pick which billing team to charge per project. When you do run out of credits, the CLI points you straight at the top-up page instead of failing with a generic error. The release also dedupes duplicate models from OpenAI-compatible upstreams, simplifies the OpenClaw installer, and fixes a collision in shared `AGENTS.md` / `CLAUDE.md` rulesHigh5/14/2026
v3.12.0ByteRover CLI 3.12.0 turns the Tasks page in `brv webui` into a per-project history. Every curate, query, and review you run is now saved across daemon restarts, with filters, pagination, bulk delete, and provider/model attribution. The release also fixes a sync regression that caused `brv push` to ship dangling `related:` references, and tidies up frontmatter so synthesized files render properly in the dashboard. ## What's Changed ### โœจ Added * **Tasks history in `brv webui`.** Past cuHigh5/9/2026
v3.10.3This release makes daemon-version mismatches visible. After you upgrade `brv`, the background daemon can keep running on the older version with no obvious way to tell. The REPL header and MCP tool replies now flag the drift so you know to run `brv restart`. ## What's Changed ### ๐Ÿ› Fixed * **No more silent daemon-version mismatches.** After upgrading `brv`, the background daemon could keep running on the older version and you'd have no way to tell, leading to confusing "why doesn't thisHigh5/5/2026
v3.10.1This release cuts the wait at the end of `brv curate`, adds a per-project switch to opt out of human-in-the-loop reviews, and tightens up two error paths: `brv vc status` / `pull` now name the files in the way, and OpenAI-compatible provider setup catches bad URLs before they break the REPL. ## What's Changed ### โœจ Added * **`brv review --disable` / `--enable`, per project.** Opt a project out of the human-in-the-loop review prompts and backup writes, or turn them back on. Run `brv reviHigh4/30/2026
v3.10.0This release adds **OpenClaude as a connectable agent**, polishes the **`brv login` browser flow**, and fixes a handful of reliability and billing rough edges, including a cold-start hang, a `brv restart` glitch under nested shell wrappers, and over-counting on multi-folder `brv curate` runs. ## What's Changed ### โœจ Added * **Connect OpenClaude as an agent.** Run `brv connectors install "OpenClaude"`, or pick it from `/connectors` in the REPL. By @DatPham-6996 in #548 ### ๐Ÿ”„ Changed High4/28/2026
v3.8.3A small quality-of-life release. `brv login` now opens your browser by default, the provider picker tells you what each provider does, and `brv webui` works correctly when `brv` is installed under a hidden folder like `~/.nvm` or `~/.asdf`. ## ๐ŸŒŸ Highlights ### ๐Ÿ” Browser sign-in by default Just run `brv login` and sign in in your browser. On CI or a remote shell, use `brv login --api-key <key>` instead. SSH and non-interactive shells are detected automatically and skip the browser, so High4/23/2026
v3.8.2This release fixes a styling bug in `brv webui`. If you upgraded to v3.8.0 or v3.8.1 and the web UI looked broken, this release makes it look right again. ## ๐ŸŒŸ Highlights ### ๐ŸŽจ Web UI styles now render correctly from the installed CLI After installing with `curl -fsSL https://byterover.dev/install.sh | sh`, opening `brv webui` showed dialogs and side sheets unstyled: no backdrop, wrong position, missing borders. Still usable, but clearly broken. v3.8.2 fixes this. The web UI now loHigh4/22/2026
v3.7.1This patch release **cleans up `brv vc` noise**: runtime ranking data no longer churns your tracked notes, and OS / editor temp files are now ignored by default. It also **refreshes the agent skill template** so connected coding agents know where to look up your past queries and curates. ## ๐ŸŒŸ Highlights ### ๐Ÿงน `brv vc status` stays clean after queries and curates Until now, every `brv search`, `brv query`, or agent curate quietly updated ranking data (things like *how often a note was High4/20/2026
v3.7.0This release adds **Intel Mac support** to the native installer so ByteRover CLI installs out of the box on `x86_64` Macs, and patches a **high-severity npm advisory** in our transitive dependencies. ## ๐ŸŒŸ Highlights ### ๐ŸŽ Intel Mac (`darwin-x64`) install support Until now, running our one-line installer on an Intel Mac stopped with an "Apple Silicon only" error. v3.7.0 publishes a `darwin-x64` tarball alongside the existing builds, so the same command now works everywhere: ```bash High4/18/2026
v3.6.1This release is a **maintenance rebuild of v3.6.0** - no code or behavior changes. v3.6.1 reships the v3.6.0 feature set with our refreshed release pipeline configuration so the distributed binary is built against the correct settings. ## Why 3.6.1? v3.6.0 was packaged before our release pipeline configuration was fully updated, so the published binary shipped with stale build settings. v3.6.1 is a clean rebuild from the same source with the correct configuration baked in. **If you instHigh4/18/2026
v3.6.0This release introduces **`brv dream`** โ€” an idle-time cleanup that keeps your context tree tidy โ€” and **`brv query-log`**, so you can finally see what you've asked and how well it was answered. ## ๐ŸŒŸ Highlights ### ๐ŸŒ™ `brv dream` โ€” tidy up your context tree while you're away Over time, your context tree collects duplicates, stale drafts, and cross-topic connections that never quite got written down. `brv dream` cleans it up in the background so you don't have to. When it runs, it doHigh4/17/2026
v3.5.1A small patch that improves how ByteRover's MCP tools communicate with coding agents. ## ๐ŸŒŸ Highlights ### ๐Ÿ”ง Smarter MCP tool descriptions The `cwd` parameter on `brv-curate` and `brv-query` MCP tools now clearly tells the calling agent when a project path is required (Claude Desktop, hosted MCP) vs. optional (Cursor, Cline, Zed, Claude Code). This reduces failed tool calls from agents that previously omitted or guessed the path. No action needed on your side โ€” just upgrade and `brvHigh4/16/2026
v3.5.0This release adds **Claude Desktop** as a first-class connector, cleans up how the installer handles its bundled Node.js, and patches a security advisory. ## ๐ŸŒŸ Highlights ### ๐Ÿ–ฅ Claude Desktop โ€” connect ByteRover to the Claude Desktop app If you use the Claude Desktop app, you can now plug ByteRover into it so Claude can search and save into your context tree directly โ€” the same memory you already use from `brv`. One command to set it up: - **`brv connectors install "Claude DesktopHigh4/15/2026
v3.4.0This release introduces **`brv swarm`** โ€” a single place to search and save across all your knowledge tools (ByteRover context tree, Obsidian, GBrain, OpenClaw Memory Wiki, and your local notes) โ€” plus an important security update and a few cleanups that make `brv vc` and knowledge search feel tidier. ## ๐ŸŒŸ Highlights ### ๐Ÿง  `brv swarm` โ€” one place to search all your notes If you keep knowledge in more than one tool โ€” an Obsidian vault, a folder of markdown notes, a team wiki, plus your BHigh4/14/2026
v3.3.0This release adds **`brv search`** - a pure BM25 retrieval command that ranks context-tree results with zero LLM calls, zero provider config, and zero token cost - plus a long-requested `--timeout` flag for `brv curate` and `brv query` so slow local models can finish their work. A handful of legacy-sync papercuts get cleaned up along the way. ## ๐ŸŒŸ Highlights ### ๐Ÿ”Ž `brv search` - zero-cost BM25 over the context tree A new command that ranks files in your context tree using pure BM25 retrHigh4/11/2026
v3.2.0This release brings **multi-repo knowledge sharing** to `brv`. You can now run `brv` from monorepo subdirectories and sibling checkouts without spawning a second `.brv/`, pull context from other projects as read-only knowledge sources, and see everything the resolver figured out at a glance from `brv status`. The tagline: one project's knowledge, any worktree's workflow. ## ๐ŸŒŸ Highlights ### ๐ŸŒณ Git-style worktree links `brv worktree add [path]` registers a subdirectory or sibling checkoutMedium4/11/2026
v3.1.0This release is all about **adaptive knowledge**. `brv` now learns from how its knowledge is actually used โ€” frequently-consulted entries rise in search, large entries are stored as compressed abstracts so they cost fewer tokens to consult, and the agent quietly distills patterns and decisions from each REPL session into durable knowledge. The longer you use it, the sharper it gets. ## ๐ŸŒŸ Highlights ### ๐Ÿ“ˆ Adaptive knowledge scoring The knowledge base now ranks results by usage. A new hotMedium4/10/2026
cipher-finalLatest release: cipher-finalHigh4/8/2026

Dependencies & License Audit

Loading dependencies...

Similar Packages

mcp-verified-repo-memoryProvide reliable, repository-scoped memory for AI coding agents with code citations, just-in-time verification, and stale-proof data management.main@2026-06-02
claude-code-safety-net๐Ÿ›ก๏ธ Enhance code safety with Claude Code Safety Net, a tool designed to identify and mitigate risks in your codebase effectively.main@2026-05-31
ori-cliAgentic coding harness with persistent memory and a REPL body. Built on Ori Mnemos. Open source must win.master@2026-05-08
continueโฉ Source-controlled AI checks, enforceable in CI. Powered by the open-source Continue CLIv1.2.22-vscode
git-notes-memory๐Ÿง  Store and search your notes effectively with Git-native memory storage, enhancing productivity for Claude Code users.main@2026-06-06

More in MCP Servers

AstrBotAgentic IM Chatbot infrastructure that integrates lots of IM platforms, LLMs, plugins and AI feature, and can be your openclaw alternative. โœจ
agentscopeBuild and run agents you can see, understand and trust.
claude-plugins-officialOfficial, Anthropic-managed directory of high quality Claude Code Plugins.
langchain4jLangChain4j is an open-source Java library that simplifies the integration of LLMs into Java applications through a unified API, providing access to popular LLMs and vector databases. It makes impleme