Ramorie CLI MCP server
Commit Oct 21, 2026Provides long-term memory and task management for AI agents through a shared MCP server for tools like Claude Code and Cursor.
Also known as kutbudev/ramorie-cli
- Stars
- 1
- Tools
- 15
- Downloads, 30 days
- —
- Latest version
- —
- Last commit
- Oct 21, 2026
- License
- MIT
What is Ramorie CLI MCP?
Ramorie CLI provides long-term memory and task management for AI agents through a shared MCP server for tools like Claude Code and Cursor. It exposes 15 tools, such as setup_agent, list_projects, remember.
Summary generated from the project's README
About Ramorie CLI MCP server
Ramorie CLI provides a persistent memory and task management system designed specifically for AI agents. It solves the problem of coding agents losing context between sessions by offering a shared MCP server that tools like Claude Code, Cursor, Windsurf, and Codex can access. This allows multiple AI clients to read from and write to a single, project-scoped memory bank.
Users can manage actionable work items through a task system with priorities and status tracking, or store technical insights using semantic search. The system connects to a hosted PostgreSQL service, so no local database setup is required. For advanced features like intelligent tagging and task analysis, users can provide a Google Gemini API key. The CLI supports various installation methods including Homebrew, npm, and Go, and offers a desktop version that automates the configuration of local embedding models and AI tool detection.
What you can do with it
- 1Maintain long-term agent contextAn AI agent can store technical decisions in a shared memory so it remembers them in future sessions.
- 2Track actionable development workA developer can instruct an AI to create and organize a list of bugs or features to implement.
- 3Store and retrieve technical insightsYou can save specific coding patterns or lessons learned to be retrieved via semantic search later.
- 4Visualize project progressA user can request a visual overview of their current tasks using a kanban board.
Written from the project's README. Every use case is checked against a quote from it.
Ramorie CLI MCP tools 15
Detected from the READMEsetup_agentlist_projectsrememberfindrecalltaskload_skillmemoryget_statsget_agent_activitysurface_contextcreate_project
| setup_agent | — | |
| list_projects | — | |
| remember | — | |
| find | — | |
| recall | — | |
| task | — | |
| load_skill | — | |
| memory | — | |
| get_stats | — | |
| get_agent_activity | — | |
| surface_context | — | |
| create_project | — |
Show all 15 toolsShow fewer tools
manage_subtasksentityadmin
| manage_subtasks | — | |
| entity | — | |
| admin | — |
Configuration
GEMINI_API_KEY
| Name | Description | |
|---|---|---|
| GEMINI_API_KEY | — |
Automated checks
Ran Nov 14, 2026Full report →Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points
- Source repository linked and reachablekutbudev/ramorie-cli is public and reachablegithub.com/kutbudev/ramorie-cli ↗Pass
- Package names this repository as its sourceNot checked
- Build provenance (Sigstore attestation)Not checked
- Open source license (OSI)MITspdx.org/licenses/MIT.html ↗Pass
- Commit or release in the last yearLast commit or release 2026-10-21github.com/kutbudev/ramorie-cli ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNot checked
- No critical advisory in the package or its dependenciesNot checked
- No hidden instructions to the model15 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNo tool named like delete, drop or execPass
- Remote endpoint answers the MCP handshakeNot checked
README
View on GitHub✨ What is Ramorie?
Ramorie gives AI agents long-term memory. Coding agents forget everything when a session ends; Ramorie is an MCP memory server they can write decisions to and search again later, so Claude Code, Cursor, Codex, Windsurf, VS Code and other MCP clients share one persistent, project-scoped memory.
Want it on your Mac with zero setup? Ramorie Desktop bundles the database, a local embedding model and this MCP server, detects your AI tools and configures them for you. It is free for 6 months. See Long-term memory for AI agents.
The CLI provides:
- 🎯 Smart Task Management — Create, organize, and track tasks with priorities, tags, and progress
- 🧠 Memory System — Store and retrieve knowledge, insights, and learnings with semantic search
- 🤖 AI Integration — Gemini-powered suggestions, task analysis, and intelligent tagging
- 📊 Visual Dashboards — Kanban boards, burndown charts, and project statistics
- 🔗 MCP Support — Model Context Protocol integration for AI agents (Cursor, Claude, etc.)
Perfect for developers, AI agents, and anyone who wants to capture knowledge while managing tasks.
Upgrading from v6 or earlier? See
CHANGELOG.mdfor the v7.0.0 breaking-changes summary (MCP_meta.protocol_reminderenvelope, removeddecision/skill/manage_focustools,admin.switch_orgdropped). v7.1.0 adds aramorie setup hooksalias forramorie setup-hooks— either command works.
🚀 Installation
Homebrew (Recommended for macOS/Linux)
brew tap kutbudev/homebrew-tap
brew install ramorie
npm (Cross-platform)
npm install -g ramorie
Go Install
go install github.com/kutbudev/ramorie-cli/cmd/jbraincli@latest
Direct Download
Download pre-built binaries from the releases page:
| Platform | Architecture | Download |
|---|---|---|
| macOS | Apple Silicon (M1/M2/M3) | ramorie_darwin_arm64.tar.gz |
| macOS | Intel | ramorie_darwin_amd64.tar.gz |
| Linux | x86_64 | ramorie_linux_amd64.tar.gz |
| Linux | ARM64 | ramorie_linux_arm64.tar.gz |
| Windows | x86_64 | ramorie_windows_amd64.zip |
curl Install Script
curl -sSL https://raw.githubusercontent.com/kutbudev/ramorie-cli/main/install.sh | bash
Build from Source
git clone https://github.com/kutbudev/ramorie-cli.git
cd ramorie-cli
go build -o ramorie ./cmd/jbraincli
🏁 Quick Start
1. One-command full install (v7.0.0+)
ramorie setup is now a complete onboarding flow — auth + MCP install for
every detected client + Persistent Memory Protocol hooks (Claude Code,
Codex) + protocol rules-files (Cursor, Windsurf) + vault check + doctor
health-check, all in one run:
ramorie setup # full install (idempotent — re-run anytime)
ramorie doctor # health check across config / vault / MCP / hooks / rules
ramorie setup-hooks status # see protocol install state per client
The legacy interactive menu is still available via ramorie setup --legacy.
Credentials are written to ~/.ramorie/config.json.
If you only want auth without the protocol surface, the per-step commands remain:
ramorie setup login # email + password only
ramorie mcp install # MCP TUI only
ramorie setup-hooks install --client claude-code # one client at a time
2. Create Your First Project
ramorie project create "My Project"
ramorie project list
3. Start Managing Tasks
# Create a task
ramorie task create "Implement user authentication" --priority H
# View your kanban board
ramorie kanban
# Start working on a task
ramorie task start <task-id>
# Mark it complete
ramorie task complete <task-id>
4. Store Knowledge
# Remember important insights (project resolves by name, short id, or UUID)
ramorie remember "Use bcrypt with 12 rounds for password hashing" -p "My Project"
# Pipe content from stdin (great for multi-line / generated text)
echo "Use bcrypt with 12 rounds for password hashing" | ramorie remember -p "My Project"
# Comma-separated tags + JSON output for scripts/agents
cat memo.md | ramorie remember -p "ramorie-cli" -t cli,docs --json
# Search your memories
ramorie find "password"
📋 Core Features
1. Task Management
- ✅ Create, list, start, complete tasks
- ✅ Priority levels (H/M/L)
- ✅ Status tracking (TODO/IN_PROGRESS/IN_REVIEW/COMPLETED)
- ✅ Progress tracking (0-100%)
- ✅ Task annotations and notes
- ✅ Detailed task information display
- 🔄 Task dependencies (coming soon)
2. Project Organization
- ✅ Multi-project support
- ✅ Active project switching
- ✅ Project-scoped tasks and memories
3. Memory System
- ✅ Store development insights and learnings
- ✅ Search and recall memories
- ✅ Project-specific or global memory views
4. Annotation System
- ✅ Add notes to tasks
- ✅ Track implementation details
- ✅ View annotation history with timestamps
5. Visual Management
- ✅ Beautiful kanban board with priority indicators
- ✅ Terminal-width responsive design
- ✅ Real-time task counts per status
🛠 Configuration
Cloud Service
The CLI connects to a hosted PostgreSQL service automatically. No database setup required! Simply register for an account and start using the tool.
Account Setup
After installation, authenticate:
# Interactive menu (includes register, login, API key, status)
ramorie setup
# Or directly log in with an existing account
ramorie setup login
Your API key will be stored in ~/.ramorie/config.json.
Gemini API Key Setup
Some advanced features (AI-powered suggestions, tag generation, etc.) require a Google Gemini API key. Prerequisite: Register for a Gemini API key if you haven't already.
To securely set or update your Gemini API key:
ramorie config set-gemini-key
You will be prompted to enter your key, which will be stored securely in your home directory (~/.ramorie_gemini_key, permissions 0600).
To remove your Gemini API key:
ramorie config set-gemini-key --remove
Environment Variables:
If you prefer, you can set the GEMINI_API_KEY environment variable instead of using the CLI command. The CLI will prioritize the environment variable if both are set.
Data Model
Key Tables:
tasks- Task managementprojects- Project organizationmemory_items- Knowledge storagecontexts- Context groupingannotations- Task notestags- Tagging system
Commands
🔴 Essential — daily
| Command | Purpose |
|---|---|
ramorie task |
List, create, update, link, note tasks |
ramorie memory |
List, get, link memories |
ramorie project |
Manage projects (accepts name, short id, or UUID) |
ramorie remember <text> |
Quick memory create (auto-detects type, supports stdin pipe + --json) |
ramorie find <term> |
Hybrid memory search (HyDE + rerank + entity graph) |
ramorie ui |
Interactive 3-pane TUI navigator (Yazi-style) |
🟡 Common — frequent
| Command | Purpose |
|---|---|
ramorie kanban -p <project> |
Beautified three-column board |
ramorie stats |
Task counts (todo / in-progress / done) — auto-JSON when piped |
ramorie activity [--burndown] |
Activity feed or burndown report — auto-JSON when piped |
ramorie subtask |
Manage subtasks |
ramorie context |
Manage contexts and packs |
🟢 Admin — setup
| Command | Purpose |
|---|---|
ramorie setup |
Interactive auth + vault unlock|lock|status subgroup |
ramorie unlock |
Unlock the encrypted vault (alias of setup vault unlock, since v6.3.5) |
ramorie lock |
Lock the encrypted vault (alias of setup vault lock, since v6.3.5) |
ramorie config |
Show config, set API key, set Gemini key |
ramorie mcp |
MCP server management |
ramorie hook |
Claude Code PreToolUse hook |
ramorie org |
Organizations + vault + encryption subgroups |
Project / org / task arguments accept name, 8-char short ID, or full UUID.
List commands (
task list,memory list,activity) render as responsive lipgloss tables. Default order is chronological ascending (newest at the bottom), with a"N of M · ↓ newest"subtitle. Use--limit Nto cap rows and--newest-firstto flip the order.
🤖 AI Agent Decision Guide
When to use task create vs remember
Use task create for:
- ✅ Actionable work items that need to be completed
- ✅ Future tasks you need to track and execute
- ✅ Bugs to fix or features to implement
- ✅ Work that has clear completion criteria
# Examples of GOOD task create usage:
ramorie task create "Fix authentication bug in login endpoint" --priority H
ramorie task create "Implement user profile editing feature"
ramorie task create "Write unit tests for payment module"
ramorie task create "Deploy version 2.1 to production"
Use remember for:
- ✅ Insights and learnings from completed work
- ✅ Technical solutions you discovered
- ✅ Best practices and patterns
- ✅ Things to avoid or lessons learned
- ✅ Knowledge that will help future development
# Examples of GOOD remember usage:
ramorie remember "OAuth requires redirect_uri to match exactly - case sensitive" -p "My Project"
ramorie remember "Use bcrypt with 12 rounds for password hashing" -p "My Project" -t auth,security
echo "Redis connection pooling reduces latency by 40%" | ramorie remember -p "My Project"
cat lessons.md | ramorie remember -p "My Project" -t perf --json
-paccepts project name, 8-char short ID, or full UUID. Content can come from positional args or piped stdin (v6.4.0+).
Use task note for:
- ✅ Progress updates on existing tasks
- ✅ Implementation details and decisions
- ✅ Blocking issues or dependencies
- ✅ Code snippets or specific technical notes
# Examples of GOOD task note usage:
ramorie task note a1b2c3d4 "Switched from JWT to session-based auth for better security"
ramorie task note a1b2c3d4 "Blocked: waiting for API key from third-party service"
ramorie task note a1b2c3d4 "Performance improved 3x after adding database indexes"
🎯 Workflow Examples
Starting a New Feature
# 1. Make sure the project exists (projects are passed per-command via --project)
ramorie project list
# 2. Create high-priority task
ramorie task create "Implement OAuth integration" --project my-app --priority H --tags "auth,api"
# 3. Start working
ramorie task start <task-id>
# 4. View progress
ramorie kanban
# 5. Update progress as you work
ramorie task progress <task-id> 50
# 6. Store learnings
ramorie remember "OAuth requires redirect_uri to match exactly - case sensitive" -p my-app
# 7. Complete task
ramorie task complete <task-id>
Daily Standup Prep
# Check kanban for current status
ramorie kanban
# List your in-progress tasks
ramorie task list --status IN_PROGRESS
# Review recent memories for insights
ramorie memory list | head -10
# Check completed tasks
ramorie task list --status COMPLETED
Project Retrospective
# Review all project memories
ramorie memory list
# Check task completion stats via kanban
ramorie kanban
# Search for specific learnings
ramorie find "performance"
ramorie find "bug"
🔧 Advanced Usage
Cross-Project Memory Access
# View memories from all projects
ramorie memory list --all
# Search across all projects
ramorie find "database" --all
Task Dependencies & Workflows
# Create linked tasks with context
ramorie task create "Backend API" --priority H --context "feature-x"
ramorie task create "Frontend UI" --priority M --context "feature-x"
ramorie task create "Integration tests" --priority L --context "feature-x"
Bulk Operations
# List tasks by multiple criteria
ramorie task list --priority H --status TODO --context "urgent"
# Filter memories by search
ramorie find "TypeScript" --all
📊 Data Model
The CLI uses a PostgreSQL database with the following key relationships:
- Projects → contain Tasks and Memory Items
- Tasks → have Annotations and Tags
- Memory Items → have Tags and can link to Tasks
- Contexts → group related Tasks and Memories
All data is persisted and synchronized across CLI sessions.
🚨 Status Codes & Priorities
Task Status
TODO- Not startedIN_PROGRESS- Currently workingIN_REVIEW- Awaiting review/QACOMPLETED- Finished
Priority Levels
- 🔴
H(High) - Urgent/Critical - 🟡
M(Medium) - Normal priority - 🟢
L(Low) - Nice to have
💡 AI Agent Best Practices
For Task Management:
- Always specify project: Use
--projectflag to specify which project to operate on - Use descriptive task names: Include what, not how
- Track progress with annotations: Document blockers, decisions, and progress
- Use task info for context: Before working on a task, review its full details
- Complete tasks promptly: Mark done when finished to maintain accurate status
For Knowledge Management:
- Store insights immediately: Use
ramorie rememberto capture learnings as they happen - Be specific in memories: Include context, not just solutions
- Search before creating: Check existing memories and tasks to avoid duplication
- Use annotations for implementation details: Keep task-specific notes with the task
- Separate concerns: Use tasks for work items, memories for knowledge, annotations for progress
For Workflow Optimization:
- Start with kanban overview:
ramorie kanbangives complete project status - Use partial UUIDs: First 8 characters are sufficient for task operations (e.g.,
a1b2c3d4) - Review task info before starting: Understand context and previous annotations
- Document as you work: Add annotations during development, not just at the end
- Capture learnings in the moment: Don't wait until the end to record insights
Common Anti-Patterns to Avoid:
❌ Creating tasks for already completed work ❌ Using remember for future work items ❌ Storing implementation details in memories instead of task annotations ❌ Creating duplicate tasks without checking existing ones ❌ Forgetting to mark tasks as complete
🔌 Integration
The CLI integrates with:
- Cloud PostgreSQL - Hosted data storage (no setup required)
- API Authentication - Secure API key-based access
- Terminal - Full CLI interface with colors and tables
- Cross-platform - Works on macOS, Linux, and Windows
This tool is designed for developers and AI agents who want powerful task management with persistent memory storage, all accessible through a beautiful command-line interface.
🤖 MCP Integration (AI Agents)
Ramorie CLI includes a built-in Model Context Protocol (MCP) server, making it compatible with AI coding assistants like Cursor, Claude Desktop, and other MCP-enabled tools.
Setup for Cursor/Claude
Add to your MCP configuration:
{
"mcpServers": {
"ramorie": {
"command": "ramorie",
"args": ["mcp", "serve"]
}
}
}
Available MCP Tools (v6.9.0 — 15 tools)
| Tool | Tier | Description |
|---|---|---|
setup_agent |
Core | Initialize session, auto-detect project from cwd |
list_projects |
Core | List personal + org projects |
remember |
Core | Store memory (auto type detection; todo: prefix → task) |
find |
Core | Hybrid semantic + lexical search (HyDE + rerank) |
recall |
Core | FTS search; precision: true routes to find |
task |
Core | Unified task ops: list/get/create/start/complete/stop/progress/note/move |
load_skill |
Core | Render a skill memory as Claude Code-format markdown (frontmatter + body) |
memory |
Common | Unified memory ops + skill generation via goal |
get_stats |
Common | Task statistics |
get_agent_activity |
Common | Agent timeline query |
surface_context |
Common | File/domain-scoped context surfacing |
create_project |
Advanced | Create project |
manage_subtasks |
Advanced | Subtask CRUD |
entity |
Advanced | Knowledge graph ops (10 actions) |
admin |
Advanced | consolidate/cleanup/orgs/export/import/plan/analyze |
v5.0.0 upgrade note: Legacy tools (
create_task,add_memory,search_memories,create_decision,activate_context_pack,skill,decision,get_active_task,manage_focus) have been removed. Restart your MCP client (Claude Code / Cursor / Windsurf) after upgrading. Runramorie mcp toolsfor the live list.
📊 Reports & Analytics
# Project statistics
ramorie stats
# Activity feed (last 7 days)
ramorie activity -d 7
# Burndown report
ramorie activity --burndown
🔧 Configuration
API Key Storage
Your credentials are stored securely in ~/.ramorie/config.json.
Gemini AI Setup (Optional)
For AI-powered features (suggestions, analysis, auto-tagging):
ramorie config set-gemini-key
Or set the environment variable:
export GEMINI_API_KEY="your-api-key"
🤖 MCP Integration (AI Agents)
Ramorie includes a built-in MCP (Model Context Protocol) server, allowing AI agents like Cursor, Windsurf, Claude Desktop, and others to manage tasks and memories directly.
Quick Setup
Run this command to get your MCP configuration:
ramorie mcp config
# or for Codex CLI users
ramorie mcp config --client codex
Windsurf / Cursor Configuration
Add the following to your MCP config file:
Windsurf: ~/.codeium/windsurf/mcp_config.json
Cursor: ~/.cursor/mcp.json
If installed via Homebrew (macOS/Linux):
{
"mcpServers": {
"ramorie": {
"command": "ramorie",
"args": ["mcp", "serve"]
}
}
}
If installed via npm:
{
"mcpServers": {
"ramorie": {
"command": "npx",
"args": ["-y", "@ramorie/cli", "mcp", "serve"]
}
}
}
If installed via Go:
{
"mcpServers": {
"ramorie": {
"command": "ramorie",
"args": ["mcp", "serve"]
}
}
}
Note: Make sure
ramorieis in your PATH, or use the full path (e.g.,/opt/homebrew/bin/ramorieor~/.local/bin/ramorie).
Codex CLI Configuration
Codex CLI uses ~/.codex/config.toml. Generate a ready-to-paste snippet with:
ramorie mcp config --client codex
Then append the output to your config file:
[mcp_servers.ramorie]
command = "ramorie"
args = ["mcp", "serve"]
enabled = true
Restart Codex CLI (or the Codex app) after editing so it reloads the MCP settings.
Available MCP Tools
See the full 15-tool table above (v6.9.0). For a live list straight from the binary:
ramorie mcp tools
Verify MCP Server
# List all available tools
ramorie mcp tools
# Start MCP server manually (for testing)
ramorie mcp serve
📝 Quick Reference
# Essential Commands
ramorie setup # Authenticate (email + password)
ramorie project list # List all projects
ramorie kanban # Visual task board
ramorie task create "Description" # New task
ramorie task show <id> # Task details
ramorie task start <id> # Begin working
ramorie task note <id> "Note" # Add progress note
ramorie remember "Insight" -p <project> # Store knowledge (or pipe via stdin)
ramorie find "term" # Hybrid memory search
ramorie unlock # Unlock encrypted vault (alias of `setup vault unlock`)
ramorie lock # Lock encrypted vault (alias of `setup vault lock`)
ramorie task complete <id> # Complete task
# Decision Guide
# Need to do work? → task create
# Making progress? → task note
# Learned something? → remember
# Need overview? → kanban
# Need details? → task show
🌐 Links
- Website: ramorie.com
- Documentation: ramorie.com/docs
- Releases: GitHub Releases
- Homebrew Tap: kutbudev/homebrew-tap
📄 License
MIT License - see LICENSE for details.
Ramorie CLI MCP questions
What tools does Ramorie CLI MCP provide?
It exposes 15 tools, including setup_agent, list_projects, remember, find, recall.
What is Ramorie?
Ramorie gives AI agents long-term memory, allowing them to write decisions to and search them later via an MCP memory server.
Does it require a database setup?
No, the CLI connects to a hosted PostgreSQL service automatically.