mako-ai — Claude Code plugin: what it does, how to install
Teaches Claude Code when to query Mako MCP tools for code and database schemas instead of grep-walking repositories.
/plugin marketplace add anthropics/claude-plugins-community
/plugin install mako-ai@claude-communityWhat mako-ai does
This plugin connects Claude Code to the Mako project intelligence server. Its .mcp.json file starts npx -y agentmako mcp over stdio, helping the model pick structured context packets and diagnostics instead of running raw grep across files.
The package adds eight skills under the mako-ai namespace. These include the entry point /mako-ai:mako-guide and dedicated assistants such as /mako-ai:mako-database or /mako-ai:mako-trace. Tool definitions are fetched dynamically through MCP ToolSearch.
Node.js 20 or newer must be available on PATH. Without an existing Mako project initialized or attached in the repository, the tools will not work.
What it helps with
- replace brute-force grep searches with structured context packets
- inspect database schemas and verify RLS audit rules
- trace execution paths and code neighborhoods
- configure the agentmako stdio MCP server for Claude Code
Example prompts for the agent
- /mako-ai:mako-guide explain how to navigate this codebase
- /mako-ai:mako-database audit current schema and RLS policies
- /mako-ai:mako-trace track call chains in the auth module
How to install mako-ai
Plugin marketplace
/plugin marketplace add anthropics/claude-plugins-community/plugin install mako-ai@claude-communityStep by step, with commands for each platform: Claude Code plugins: marketplaces, installing, and building your own
Questions about mako-ai
How do I install mako-ai in Claude Code?
Run both commands in sequence: first /plugin marketplace add anthropics/claude-plugins-community, then /plugin install mako-ai@claude-community.
What does mako-ai do?
It packages guidance skills and wires Claude Code to the agentmako MCP server for structural code navigation, graph lookups, and schema checks.
What are the prerequisites for mako-ai?
You need Node.js 20+ on your PATH so npx can launch the server. The target repository must also have an initialized or attached Mako project.
Source text
mako-ai-claude-plugin · Apache-2.0
# Mako AI Claude Code Plugin This plugin packages Mako guidance and MCP wiring for Claude Code. It does not replace Mako's typed MCP tools. It teaches Claude Code when to use them, while `.mcp.json` starts the existing `agentmako mcp` stdio server. ## Prerequisites - Claude Code installed. - Node.js 20+ on `PATH` (the plugin's `.mcp.json` runs `npx -y agentmako mcp`, which fetches and caches the published `agentmako` package automatically). - A Mako project already attached or initialized for the repo you are using. ## Local Development From the Mako repo root: ```powershell claude plugin validate .\mako-ai-claude-plugin claude --plugin-dir .\mako-ai-claude-plugin ``` Inside Claude Code, the skills appear under the `mako-ai` namespace: - `/mako-ai:mako-guide` - `/mako-ai:mako-discovery` - `/mako-ai:mako-trace` - `/mako-ai:mako-neighborhoods` - `/mako-ai:mako-graph` - `/mako-ai:mako-database` - `/mako-ai:mako-code-intel` - `/mako-ai:mako-workflow` Use `/mcp` in Claude Code to verify the plugin-provided `mako-ai` MCP server is connected. ## Skill Layout `mako-guide` is the entry skill and owns cross-cutting feedback policy. The remaining skills are category-scoped so Claude Code can load only the guidance relevant to the user's intent. Live tool schemas come from MCP ToolSearch and `tools/list`. The skills should not be treated as schema references. ## Feedback Policy Use `agent_feedback` selectively, not after every call. First use `recall_tool_runs` to find the prior run and copy its `requestId` into `agent_feedback.referencedRequestId`. Use `finding_ack` only for reviewed static findings from tools such as `ast_find_pattern` and `lint_files`; do not use it to rate tool usefulness.
© drhalto · Apache-2.0 · drhalto/agentmako
