OpenReflex
Local project memory for coding agents that suggests paths and flags stalled work
- Pricing
- Open source
- Platforms
- CLI, self-hosted, macOS
- Runs
- Locally
- Version
- 0.10.2
- Listing updated
- 2 days ago
About OpenReflex
OpenReflex is an open-source MCP server for developer tools. It runs locally on your machine and is started by your AI client.
This is the optional diagnostics side of OpenReflex. The plugin itself works through lifecycle hooks in Claude Code, Codex, Cursor and OpenCode, so the model does not need to call it. The MCP server exists for inspecting what the memory has learned without loading it into normal agent context.
The README describes how it works: hooks record how tasks go, repeated behaviour is compiled into project-scoped procedures called Reflexes, and relevant ones are injected into new tasks. A path check only names a better approach when comparable past tasks back it up. It raises a single alert when it sees repeated failing commands, identical retries or stalled progress, suggesting whether to continue, pivot or stop.
Data stays in a local SQLite database, with no account or hosted service. Only coarse project-relative metadata is stored, never file contents, commands, tool output or transcripts, and capture stays off until you approve a project. It needs Python 3.11 or newer.
Install
Run in a terminal
claude mcp add openreflex -- uvx openreflexAdd to claude_desktop_config.json
{
"mcpServers": {
"openreflex": {
"command": "uvx",
"args": [
"openreflex"
]
}
}
}Add to ~/.cursor/mcp.json
{
"mcpServers": {
"openreflex": {
"command": "uvx",
"args": [
"openreflex"
]
}
}
}Add to .vscode/mcp.json
{
"servers": {
"openreflex": {
"type": "stdio",
"command": "uvx",
"args": [
"openreflex"
]
}
}
}Add to ~/.codeium/windsurf/mcp_config.json
{
"mcpServers": {
"openreflex": {
"command": "uvx",
"args": [
"openreflex"
]
}
}
}Add to your client's MCP settings (most use this shape)
{
"mcpServers": {
"openreflex": {
"command": "uvx",
"args": [
"openreflex"
]
}
}
}What it does
- Ambient project memory
- Hooks record how work went and reuse it in later tasks without the model asking.
- Path recommendations
- Compares past approaches by success, time, tool calls and context cost.
- Continue, pivot or stop
- Raises one alert on loops, stalled progress or runaway context.
- Local only storage
- Keeps a SQLite database on your machine with no account.
- Opt-in per project
- Capture stays off until you approve each project.
- Diagnostics over MCP
- Optional server for inspecting memory without bloating agent context.
Good for
- Stop an agent re-reading the same files and retrying failing commands
- Reuse the approach that worked for a similar past bug
- Preview the context a new task would receive
- Inspect what OpenReflex has learned about a project
Pricing
Open source
MIT licensed
Prices change. Check openreflex.cc before you buy.
OpenReflex FAQ
Does it read my files or transcripts?
No. The README says only coarse, project-relative metadata is stored, never contents, commands, tool output or transcripts.
Which agents does it support?
Claude Code, Codex, Cursor and OpenCode.
What does it need?
Python 3.11 or newer, installed through the Claude plugin or pipx or uv.
Is it free?
Yes, it is MIT licensed.
Is OpenReflex free?
Yes. OpenReflex is open source, so you can use and self-host it for free.
How do I add OpenReflex to Claude Code?
Run this command in your terminal: claude mcp add openreflex -- uvx openreflex
What are the best alternatives to OpenReflex?
The closest alternatives to OpenReflex are REA, Codebase Memory, GitHub, Serena MCP: the IDE for your agent and XERJ.
Alternatives to OpenReflex

REA
MCP serverCLI and MCP server that lets coding agents reverse engineer apps and binaries
Runs locally- Open source
- 464K GitHub stars
- Developer tools

Codebase Memory
MCP serverLocal MCP server that builds a code knowledge graph for AI coding agents
Runs locally- Open source
- 48.7K GitHub stars
- Knowledge and memory

GitHub
MCP serverOfficial MCP server connecting AI assistants to GitHub repos, issues and workflows
Not checked yet- Remote
- Open source
- 34.5K GitHub stars
- Developer tools

Serena MCP: the IDE for your agent
MCP serverMCP toolkit giving coding agents symbol-level code retrieval and editing
Runs locally- Open source
- 31.5K GitHub stars
- Developer tools

XERJ
MCP serverLocal search engine for AI agents with cited answers over your code and docs
Runs locally- Open source
- 19.2K GitHub stars
- Knowledge and memory

Ouroboros
MCP serverSpec-first layer that pins requirements and verifies results of AI coding agent work
Runs locally- Open source
- 6.4K GitHub stars
- Developer tools