Skip to main content
Give your Hermes agents persistent memory with ByteRover. The ByteRover memory provider automatically native retrieves relevant knowledge before each response, curates new insights after each turn, and preserves context before compression — so your agents remember decisions, patterns, and fixes across sessions.

​
Prerequisites

Ensure the following are installed and available in your PATH:
RequirementHow to install
ByteRover CLI (brv)ByteRover installtion guide
Hermes Agent (hermes)Hermes installtion guide

​
ByteRover CLI initialization

ByteRover CLI needs an LLM provider to power its curation and query features. Supported list: ByteRover, OpenRouter, Anthropic, OpenAI, Google Gemini, xAI, Groq, Mistral, DeepInfra, Cohere, Together AI, Perplexity, Cerebras, Vercel, MiniMax, GLM, Moonshot AI, and OpenAI Compatible. To see all available providers: 
brv providers list
Run one of the following to get started: 
brv providers connect byterover                                                          # Free, requires login with ByteRover account
brv providers connect anthropic --api-key sk-ant-... --model claude-sonnet-4-5-20250929  # Use your own API key: OpenAI, Anthropic, etc.

​
Hermes integration guide

Option 1 — Interactive setup (recommended): 
hermes memory setup
When prompted, select byterover from the list of available memory providers. Option 2 — Manual configuration: 
hermes config set memory.provider byterover
After memory configuration, restart your Hermes gateway: 
hermes gateway restart
If you get stuck on the gateway restart command, reinstall the gateway: 
hermes gateway install
Then start the Hermes gateway: 
hermes gateway start
Then restart the Hermes gateway: 
hermes gateway restart

​
Features

​
Context Retrieval

Before each LLM call, the provider runs brv query with the user’s message and injects the results as additional context. Your agent automatically has relevant memories, past decisions, and patterns available without any manual prompting.

​
Automatic Curation

After each conversation turn, the provider runs brv curate in the background to extract and store valuable knowledge — architectural decisions, bug fixes, recurring patterns, and more. This happens asynchronously and does not add latency to responses.

​
Pre-Compression Flush

When the agent’s context window approaches compression, the provider extracts and stores insights before the conversation history is summarized. Nothing valuable is lost when the context is compacted.

​
Hermes Agent Tools

ByteRover integration exposes three tools that your agent can call directly:
ToolDescription
brv_querySearch the persistent knowledge tree for memories, decisions, and patterns from previous sessions
brv_curateStore a fact, decision, or pattern into the knowledge tree — automatically categorized and organized
brv_statusCheck the ByteRover CLI version, tree stats, and cloud sync state
Context-tree directory: $HERMES_HOME/byterover/.brv (profile-scoped, created automatically on first use)

LLM Providers

Connect an external provider or use the built-in LLM

Onboard Context

Learn how to seed your context tree with existing knowledge

Reference

Troubleshooting, error

Local & Cloud

Exploring local & cloud options