canny
Canny MCP Server for Canny API integration with AI Assistants. Features Jira linking, batch operations, ETA management, company MRR tracking, and PM-focused workflows.
claude mcp add --transport stdio opensourceops-canny-mcp-server npx -y @opensourceops/canny-mcp \ --env CANNY_API_KEY="Your Canny API key" \ --env CANNY_SUBDOMAIN="Your Canny subdomain, e.g. <subdomain>.canny.io" \ --env CANNY_TOOL_MODE="readonly | all | discovery,posts (optional)" \ --env CANNY_DEFAULT_BOARD="Default board ID (optional)"
How to use
This MCP server exposes the Canny feedback management API as an MCP-compatible set of tools. It provides 37 tools covering boards, posts, comments, votes, users, categories, tags, changelog, status changes, ideas, insights, groups, opportunities, and Jira integration. By default the server runs in read-only mode, enabling a safe, side-effect-free workflow; you can enable write capabilities by configuring CANNY_TOOL_MODE. To use it, connect an MCP client (such as Claude Code or other MCP-enabled clients) to the server and begin issuing tool calls like canny_list_posts, canny_get_post, canny_create_post, or canny_add_vote. The tools are designed to streamline product feedback processes, support Jira integrations, and provide lightweight pagination and batch operations for efficiency. Start with the Quick Start instructions to wire up credentials, then query the available tools to confirm 37 capabilities are present. The Claude Code integration example demonstrates how to register the MCP server and start using the toolset via the Claude workflow environment.
How to install
Prerequisites:
- Node.js v20.9+, v22+, or v24+ (LTS)
- npm v9+
- MCP Client (Claude Code, Continue.dev, or any MCP-compatible client)
- Canny API Key
Installation steps:
- Install the MCP server package globally (preferred for quick start):
npm install -g @opensourceops/canny-mcp
- Create or update your MCP client configuration to use the canny-mcp-server executable (instead of npx):
{
"mcpServers": {
"canny": {
"command": "canny-mcp-server",
"env": {
"CANNY_API_KEY": "your_api_key",
"CANNY_SUBDOMAIN": "your_subdomain",
"CANNY_DEFAULT_BOARD": "your_board_id",
"CANNY_TOOL_MODE": "readonly"
}
}
}
}
- If using via npx (Quick Start):
claude mcp add --transport stdio canny \
--scope user \
--env CANNY_SUBDOMAIN=<SUBDOMAIN> \
--env CANNY_API_KEY=<API_KEY> \
--env CANNY_DEFAULT_BOARD=<BOARD_ID> \
--env CANNY_TOOL_MODE=readonly \
-- npx -y @opensourceops/canny-mcp
- Restart your IDE or environment (e.g., Claude Code) if required to load the new server. Then verify by listing tools or asking the MCP client to list available tools.
Additional notes
Notes and tips:
- By default, the server runs in readonly mode; to enable write capabilities, set CANNY_TOOL_MODE to all or configure specific tool sets as described in the docs.
- Ensure your Canny API key and subdomain are kept secure and not committed to public repos.
- The server supports batch updates and smart pagination for efficient data retrieval.
- If you encounter authentication errors, double-check that CANNY_API_KEY and CANNY_SUBDOMAIN are correctly set and that the associated board IDs exist.
- When changing configuration (e.g., tool mode), rebuild or restart the server as required by your hosting setup.
Related MCP Servers
everything-claude-code
The agent harness performance optimization system. Skills, instincts, memory, security, and research-first development for Claude Code, Codex, Cowork, and beyond.
mcp-fusion
MCP Fusion - The framework for AI-native MCP servers.
claude-code-open
Open source AI coding platform with Web IDE, multi-agent system, 37+ tools, MCP protocol. MIT licensed.
mcp-rest-api
A TypeScript-based MCP server that enables testing of REST APIs through Cline. This tool allows you to test and interact with any REST API endpoints directly from your development environment.
mem0 -selfhosted
Self-hosted mem0 MCP server for Claude Code. Run a complete memory server against self-hosted Qdrant + Neo4j + Ollama while using Claude as the main LLM.
claude-vigil
🏺 An MCP server for checkpointing and file recovery in Claude Code