Tools

What are MCP Tools?

OMC provides MCP (Model Context Protocol) tools to Claude Code agents. Agents use them for state management, note storage, code analysis, and data processing.

Unlike skills that users invoke directly, tools are used internally by agents during their work.

Tool Categories

State

Tools for managing execution mode state.

state_read, state_write, state_clear, state_list_active, state_get_status

Used by execution modes like Autopilot, Ralph, and Ultrawork to save and read their current state.

Notepad

A note system that survives context compaction.

notepad_read, notepad_write_priority, notepad_write_working, notepad_write_manual, notepad_prune, notepad_stats

Important information is preserved even when pushed out of the context window during long sessions.

Project Memory

Manages per-project long-term memory.

project_memory_read, project_memory_write, project_memory_add_note, project_memory_add_directive

Maintains project structure, rules, and learned knowledge across sessions.

LSP

Code intelligence tools via Language Server Protocol.

lsp_hover, lsp_goto_definition, lsp_find_references, lsp_diagnostics, lsp_rename and 7 other tools

Enables type information, go-to-definition, find references, error diagnostics, symbol search, and more.

AST Grep

AST-based structural code search and replacement tools.

ast_grep_search, ast_grep_replace

Search and replace using code structure patterns rather than regular expressions.

Python REPL

Performs data analysis with a persistent Python REPL.

python_repl

A Python environment with state preserved across sessions.

Session Search

Searches previous local session history and transcript artifacts.

session_search

Returns session IDs, timestamps, source paths, and matching excerpts as structured JSON.

Trace

Analyzes agent flow trace data.

trace_timeline, trace_summary

trace_timeline displays hooks, keywords, skills, agents, and tools in chronological order. trace_summary aggregates hook statistics, keyword frequency, skill activations, mode transitions, and tool bottlenecks.

Shared Memory

Cross-agent shared memory for team coordination.

shared_memory_write, shared_memory_read, shared_memory_list, shared_memory_delete, shared_memory_cleanup

Enables agents to share data across team boundaries during coordinated workflows.

Skills

Internal skill management tools.

load_omc_skills_local, load_omc_skills_global, list_omc_skills

Loads and lists available OMC skills from local project and global user directories.