feat: openMF#46 Add mcp implementation for community_chatbot

community_mcp_servers/.env.example

@@ -0,0 +1,122 @@
+# LLM Provider Configuration (Required)
+
+# Choose your LLM provider: gemini, groq, openai, or lightning
+LLM_PROVIDER=gemini
+
+# API Keys (provide based on your LLM_PROVIDER choice)
+GOOGLE_API_KEY=your_google_api_key_here
+# GROQ_API_KEY=your_groq_api_key_here
+# OPENAI_API_KEY=your_openai_api_key_here
+# LIGHTNING_API_KEY=your_lightning_api_key_here
+
+# Optional: Custom model name
+# MODEL=models/gemini-2.0-flash-exp
+
+# Optional: Lightning AI custom endpoint
+# LIGHTNING_BASE_URL=https://lightning.ai/api/v1
+
+# GitHub Agent Configuration
+
+# Required: GitHub Personal Access Token
+# Create at: https://github.com/settings/tokens
+GITHUB_PERSONAL_ACCESS_TOKEN=your_github_token_here
+
+# Optional: GitHub MCP Server Configuration
+# GITHUB_MCP_BEARER_TOKEN=your_bearer_token
+# GITHUB_MCP_SERVER_URL=https://api.githubcopilot.com/mcp/
+
+# Slack Agent Configuration
+# See: https://github.com/korotovsky/slack-mcp-server
+
+# Authentication: Choose ONE of these three methods:
+
+# Method 1: User OAuth Token (Recommended - full access)
+# Create at: https://api.slack.com/apps
+SLACK_MCP_XOXP_TOKEN=xoxp-your-token-here
+
+# Method 2: Bot Token (Limited access - invited channels only, no search)
+# SLACK_MCP_XOXB_TOKEN=xoxb-your-bot-token-here
+
+# Method 3: Browser Tokens (Stealth mode - no permissions needed)
+# Extract from browser DevTools while logged into Slack
+# SLACK_MCP_XOXC_TOKEN=xoxc-your-browser-token-here
+# SLACK_MCP_XOXD_TOKEN=xoxd-your-browser-cookie-here
+
+# Optional: Enable message posting
+# Options: empty/false (disabled), true (all channels), or comma-separated channel IDs
+# Use ! prefix to allow all EXCEPT specified channels: !C123,C456
+# SLACK_MCP_ADD_MESSAGE_TOOL=false
+
+# Optional: Auto-mark sent messages as read
+# SLACK_MCP_ADD_MESSAGE_MARK=true
+
+# Optional: Transport Configuration
+# SLACK_MCP_TRANSPORT=stdio
+# SLACK_MCP_SERVER_URL=http://127.0.0.1:13080/sse
+# SLACK_MCP_API_KEY=your-api-key-here
+# SLACK_MCP_HOST=127.0.0.1
+# SLACK_MCP_PORT=13080
+
+# Optional: Enterprise Slack settings
+# SLACK_MCP_USER_AGENT=your-browser-user-agent
+# SLACK_MCP_CUSTOM_TLS=true
+# SLACK_MCP_PROXY=http://proxy.example.com:8080
+
+# Optional: Cache configuration
+# SLACK_MCP_USERS_CACHE=/path/to/users_cache.json
+# SLACK_MCP_CHANNELS_CACHE=/path/to/channels_cache_v2.json
+
+# Optional: Logging
+# SLACK_MCP_LOG_LEVEL=info
+
+# Jira Agent Configuration
+# Uses mcp-atlassian package via uvx (Python)
+# See: https://github.com/sooperset/mcp-atlassian
+# Configured for: https://mifosforge.jira.com
+
+# Required: Jira Instance URL
+JIRA_URL=https://mifosforge.jira.com
+
+# Authentication: Choose ONE method
+
+# Method 1: API Token (Recommended for Cloud)
+# Create at: https://id.atlassian.com/manage-profile/security/api-tokens
+JIRA_USERNAME=your.email@company.com
+JIRA_API_TOKEN=your_jira_api_token_here
+
+# Method 2: Personal Access Token (Server/Data Center)
+# Go to: Profile → Personal Access Tokens → Create token
+# JIRA_PERSONAL_TOKEN=your_personal_access_token_here
+
+# Optional: SSL verification (set to false for self-signed certs)
+# JIRA_SSL_VERIFY=true
+
+# Optional: Project filtering (limit to specific projects)
+# JIRA_PROJECTS_FILTER=MIFOS,FINERACT
+
+# Optional: Read-only mode (disable write operations)
+# READ_ONLY_MODE=true
+
+# Optional: Enable only specific tools
+# ENABLED_TOOLS=jira_search,jira_get_issue
+
+# Optional: Transport Configuration
+# Default: stdio (uses uvx mcp-atlassian)
+# JIRA_MCP_TRANSPORT=stdio
+
+# Optional: Use Docker instead of uvx
+# JIRA_MCP_USE_DOCKER=false
+# JIRA_MCP_DOCKER_IMAGE=ghcr.io/sooperset/mcp-atlassian:latest
+
+# Optional: Proxy settings
+# HTTP_PROXY=http://proxy.example.com:8080
+# HTTPS_PROXY=http://proxy.example.com:8080
+# NO_PROXY=localhost,127.0.0.1
+
+# Optional: Custom HTTP headers (for corporate environments)
+# Format: key=value,key2=value2
+# JIRA_CUSTOM_HEADERS=X-Custom-Header=value
+
+# Optional: Logging
+# MCP_VERBOSE=true
+# MCP_VERY_VERBOSE=false

community_mcp_servers/.gitignore

@@ -0,0 +1,51 @@
+*.env
+
+# Python
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+.venv/
+venv/
+ENV/
+env/
+
+# uv
+uv.lock
+.python-version
+
+# IDEs
+.vscode/
+
+# Node
+node_modules/
+package-lock.json
+npm-debug.log*
+yarn-debug.log*
+yarn-error.log*
+.pnpm-debug.log*
+
+# Testing
+.pytest_cache/
+.coverage
+htmlcov/
+*.cover
+
+# Logs
+*.log
+

community_mcp_servers/README.md

@@ -0,0 +1,195 @@
+# Community AI MCP Agent
+
+A unified Python CLI for interacting with GitHub, Slack, and Jira using the Model Context Protocol (MCP) and LangChain.
+
+## Features
+
+- **GitHub Agent**: Repository management, issues, PRs, and code operations
+- **Slack Agent**: Channel management, messaging, and workspace interactions  
+- **Jira Agent**: Issue tracking, project management, and workflow automation
+- **Multi-LLM Support**: Works with Gemini, Groq, OpenAI, and Lightning AI
+- **Extensible Architecture**: Built on LangChain and LangGraph for easy customization
+
+## Quick Start
+
+### 1. Installation
+
+```bash
+# Clone the repository
+cd community_chatbot/mcp_impl
+
+# Install Python dependencies (using uv or pip)
+uv sync
+# or
+pip install -r requirements.txt
+
+# Install Node.js dependencies for Slack agent (requires Node.js)
+cd agents
+npm install
+cd ..
+```
+
+### 2. Configuration
+
+Copy the example environment file and configure your credentials:
+
+```bash
+cp .env.example .env
+```
+
+**Minimum required configuration:**
+
+```env
+# Choose your LLM provider
+LLM_PROVIDER=gemini
+GOOGLE_API_KEY=your_key_here
+
+# Enable agents as needed
+GITHUB_PERSONAL_ACCESS_TOKEN=your_token_here
+SLACK_MCP_XOXP_TOKEN=xoxp-your-token-here
+JIRA_URL=https://your-company.atlassian.net
+JIRA_USERNAME=your.email@company.com
+JIRA_API_TOKEN=your_token_here
+```
+
+### 3. Usage
+
+```bash
+# Run the GitHub agent
+python main.py github
+
+# Run the Slack agent
+python main.py slack
+
+# Run the Jira agent
+python main.py jira
+
+# Get help for any agent
+python main.py github --help
+```
+
+## Agent Details
+
+### GitHub Agent
+
+Interact with GitHub repositories using your personal access token.
+
+**Required:**
+
+- `GITHUB_PERSONAL_ACCESS_TOKEN` - [Create here](https://github.com/settings/tokens)
+
+**Capabilities:**
+
+- Repository operations (create, clone, search)
+- Issue and PR management
+- Code search and file operations
+- Workflow automation
+
+### Slack Agent
+
+Connect to Slack workspaces and manage communications.
+
+**Required:**
+
+- `SLACK_MCP_XOXP_TOKEN` - [Create Slack app](https://api.slack.com/apps)
+
+**Capabilities:**
+
+- Channel and user management
+- Message posting and retrieval
+- Workspace information
+
+**Optional:** Enable message posting with `SLACK_MCP_ADD_MESSAGE_TOOL=true`
+
+### Jira Agent
+
+Manage Jira projects and issues programmatically.
+
+**Required:**
+
+- `JIRA_URL` - Your Jira instance URL
+- `JIRA_USERNAME` - Your email
+- `JIRA_API_TOKEN` - [Create here](https://id.atlassian.com/manage-profile/security/api-tokens)
+
+**Capabilities:**
+
+- Issue CRUD operations
+- Project and sprint management
+- Advanced JQL searches
+- Custom field handling
+
+## Implementation Details
+
+- **MCP integration:** Agents use an internal MCP-based client flow implemented in [community_chatbot/mcp_impl/lib/base_agent.py](community_chatbot/mcp_impl/lib/base_agent.py). The CLI shell for each agent is created with the helper in [community_chatbot/mcp_impl/lib/base_mcp.py](community_chatbot/mcp_impl/lib/base_mcp.py) which exposes commands like `list-tools`, `chat`, `invoke-tool`, and `health`.
+
+- **MCP client used:** The code constructs a `MultiServerMCPClient` (from the `langchain_mcp_adapters` package) inside `BaseAgent.initialize()` to discover and load remote MCP tools. Tools discovered from the MCP endpoints are converted into LangGraph/LangChain-compatible tool definitions and used to create a React-style agent via `langgraph.prebuilt.create_react_agent`.
+
+- **Transport options:** Agents support multiple transport modes:
+ 	- `stdio` — runs a local process (usually an `npx` package or a Docker image) and communicates over stdio. Examples:
+  		- Slack: runs `npx slack-mcp-server --transport stdio` (see [community_chatbot/mcp_impl/agents/slack_agent.py](community_chatbot/mcp_impl/agents/slack_agent.py)).
+  		- Jira: can run `ghcr.io/sooperset/mcp-atlassian:latest` (Docker) or `npx mcp-atlassian@latest` (see [community_chatbot/mcp_impl/agents/jira_agent.py](community_chatbot/mcp_impl/agents/jira_agent.py)).
+ 	- `sse` / `streamable_http` — connects to an HTTP/SSE MCP server endpoint. Default example endpoints used by the code:
+  		- Slack HTTP default: `http://127.0.0.1:13080/sse`
+  		- Jira HTTP default: `http://127.0.0.1:8080/sse`
+  The HTTP transport builder is implemented in [community_chatbot/mcp_impl/lib/base_transport.py](community_chatbot/mcp_impl/lib/base_transport.py).
+
+- **LLM & agent creation:** The LLM provider is chosen by `LLM_PROVIDER` (see [community_chatbot/mcp_impl/lib/base_agent.py](community_chatbot/mcp_impl/lib/base_agent.py)) and the repository includes provider adapters in [community_chatbot/mcp_impl/llm_providers/](community_chatbot/mcp_impl/llm_providers/) (Gemini, Groq, Lightning). The selected LLM is passed into `create_react_agent` alongside the loaded MCP tools to form the agent executor.
+
+### GitHub Agent
+
+- `https://github.com/github/github-mcp-server` — GitHub MCP server repository.
+- `https://api.githubcopilot.com/mcp/` — default GitHub MCP endpoint used as the service URL in `agents/github_agent.py`.
+- `langchain_mcp_adapters` — MCP client package used via `MultiServerMCPClient` (see `lib/base_agent.py`).
+- `langgraph` / `langchain` — used to create the React-style agent (`langgraph.prebuilt.create_react_agent`).
+
+### Slack Agent
+
+- `https://github.com/korotovsky/slack-mcp-server` — Slack MCP server repository.
+- `npx slack-mcp-server` — npm package invoked in `agents/slack_agent.py` when using the `stdio` transport (runs `slack-mcp-server --transport stdio`).
+- `SLACK_*` environment variables (tokens) map to Slack credentials and standard Slack developer docs: <https://api.slack.com/>
+- Default local HTTP/SSE endpoint in code: `http://127.0.0.1:13080/sse` (used when `SLACK_MCP_TRANSPORT` is set to `sse`/`streamable_http`).
+
+### Jira Agent
+
+- `https://github.com/sooperset/mcp-atlassian` — Jira MCP server repository.
+- `ghcr.io/sooperset/mcp-atlassian:latest` — Docker image referenced in `agents/jira_agent.py` for the `stdio` Docker transport.
+- `npx mcp-atlassian@latest` — npm package fallback for `stdio` (non-Docker) mode.
+- Jira developer docs and API token creation: <https://id.atlassian.com/manage-profile/security/api-tokens>
+- Default local HTTP/SSE endpoint in code: `http://127.0.0.1:8080/sse` (used when `JIRA_MCP_TRANSPORT` is `sse`/`streamable_http`).
+
+## Project Structure
+
+```
+mcp_impl/
+├── agents/              # Agent implementations
+│   ├── github_agent.py
+│   ├── jira_agent.py
+│   └── slack_agent.py
+├── lib/                 # Core library
+│   ├── base_agent.py   # Base agent class
+│   ├── base_mcp.py     # MCP integration
+│   └── utils.py        # Utilities
+├── llm_providers/      # LLM provider implementations
+│   ├── gemini.py
+│   ├── groq_llm.py
+│   ├── lightning_llm.py
+│   └── __init__.py
+├── main.py             # CLI entry point
+├── .env.example        # Configuration template
+└── requirements.txt    # Python dependencies
+```
+
+## Requirements
+
+- Python >= 3.12
+- Valid API keys for your chosen LLM provider
+- Agent-specific credentials (GitHub token, Slack token, Jira credentials)
+
+## Getting API Keys
+
+- **GitHub**: [Settings → Developer settings → Personal access tokens](https://github.com/settings/tokens)
+- **Slack**: [Create a Slack app](https://api.slack.com/apps) → Install to workspace → Copy OAuth token
+- **Jira**: [Account security → API tokens](https://id.atlassian.com/manage-profile/security/api-tokens)
+- **Gemini**: [Google AI Studio](https://aistudio.google.com/app/apikey)
+- **Groq**: [Groq Console](https://console.groq.com/keys)
+- **OpenAI**: [OpenAI API Keys](https://platform.openai.com/api-keys)

community_mcp_servers/agents/__init__.py

@@ -0,0 +1,5 @@
+from .github_agent import get_github_agent
+from .jira_agent import get_jira_agent
+from .slack_agent import get_slack_agent
+
+__all__ = ["get_github_agent", "get_jira_agent", "get_slack_agent"]