tidewave-tools-usage

Tidewave MCP Tools - CRITICAL PRIORITY FOR ELIXIR DEVELOPMENT

MANDATORY: Use Tidewave MCP Tools (Skill) as Primary Interface

When working with this Elixir/Phoenix/Ash codebase, ALWAYS prioritize Tidewave MCP tools over traditional file system operations. Tidewave provides deep integration with the Elixir runtime and superior code intelligence.

Tool Usage Hierarchy

1. Code Evaluation - ALWAYS Use Tidewave

NEVER use Bash to run Elixir code! Instead use mcp__tidewave__project_eval:

• Test function behavior and debug issues
• Explore modules and their functions
• Access IEx helpers (e.g., exports(Module), h(Module.function))
• Capture IO output
• Pass arguments with the arguments parameter
• Set custom timeout for long-running operations

Example:

• ❌ WRONG: bash: mix run -e "IO.inspect(MyModule.function())"
• ✅ RIGHT: mcp__tidewave__project_eval: code: "IO.inspect(MyModule.function())"

2. Source Code Navigation - Tidewave First

Before using Grep, Glob, or Read for Elixir code:

mcp__tidewave__get_source_location - Find exact file locations instantly

• Works with: Module, Module.function, Module.function/arity
• Find dependencies: "dep:package_name"
• FASTER than grep/glob for known modules

mcp__tidewave__get_docs - Get documentation without reading files

• Module docs: "MyModule"
• Function docs: "MyModule.function/2"
• Callback docs: "c:GenServer.init/1"

Example:

• ❌ WRONG: grep: pattern: "defmodule Worker"
• ✅ RIGHT: mcp__tidewave__get_source_location: reference: "MyApp.Worker"

3. Database Operations - Direct SQL Execution

mcp__tidewave__execute_sql_query - Run SQL directly against Ecto repos

• Inspect database schema
• Query data (limited to 50 rows, use LIMIT/OFFSET for more)
• Supports parameterized queries
• Auto-detects available repositories
• Returns native Elixir data structures

mcp__tidewave__get_ecto_schemas - List all schemas and their locations

• ALWAYS use this before searching for schema files
• Returns module names with file paths