Tools Overview

Ganglia exposes tools over MCP organized into ten categories. Every tool has its own page — use the sidebar to navigate or follow the links below.

Core Reading

Tool	What it does
`code_files`	List all indexed files
`code_toc`	Table of contents (files + symbols with line numbers)
`code_grep`	Search code content by regex pattern
`code_search`	Find symbols by name or natural language
`code_get`	Read a specific function or class
`code_signature`	Read only the signature line of a function
`code_structure`	Project structure tree
`code_summary`	Full project overview
`code_quick`	Natural language query — auto-routes to the best tool

Tool	What it does
`code_callers`	Who calls this function?
`code_callees`	What does this function call?
`code_context`	Function + callers + callees in one call
`code_related`	Everything connected to a symbol (callers, callees, siblings, knowledge)
`code_trace`	Full call chain (controller → DB)
`code_imports`	Who imports this module?

Tool	What it does
`code_impact`	What breaks if I change this? (severity scored)
`code_hotspots`	Most-called, most-complex, or largest code
`code_dead`	Unreachable code with no callers
`code_diff_context`	Git diff + all affected callers
`code_changelog`	Function-level git history
`code_map`	ASCII architecture diagram
`code_error_context`	Code around a specific error line

Tool	What it does
`code_route`	HTTP route → handler mapping
`code_route_trace`	Route → controller → service → DB
`code_schema`	Database schema (Drizzle/Prisma/TypeORM)
`code_env`	Environment variable definitions and reads
`code_type`	TypeScript type/interface definition

Tool	What it does
`code_rename_symbol`	Preview all locations before renaming
`code_type_usages`	All places a type is used

Tool	What it does
`code_memory`	Persist decisions and context across sessions
`code_knowledge`	Save notes linked to code nodes in the graph
`code_annotate`	Add persistent notes to a specific function

Tool	What it does
`code_build`	Compile and map errors to graph functions
`code_test`	Run tests, map failures to functions
`code_lint`	Lint and map warnings to functions
`code_exec`	Run arbitrary commands

Requires GL_SMART_READ=true in the MCP server environment.

Tool	What it does
`code_smart_read`	LLM digest of a function (80–90% fewer tokens)
`code_smart_grep`	Search + summarized answer
`code_smart_diff`	Explain changes and impact in plain language
`code_smart_context`	Memory-first understanding — checks cache first

Tool	What it does
`code_patterns`	Auto-detect project patterns from the codebase
`code_rules`	Get project rules by scope (frontend/backend/testing/all)
`code_config`	Read and write CLAUDE.md configuration files

Tool	What it does
`zone_guard`	Intercept tool calls touching marked nodes

Tool	What it does
`code_test_for`	Find which test files cover a function
`code_projects`	List all indexed projects
`code_switch_project`	Switch active project context
`code_checkpoint`	Save or restore graph state

Index and query external documents (Markdown, CSV, YAML, JSON, plain text).

Tool	What it does
`doc_index`	Index a file into queryable chunks
`doc_list`	List all indexed documents
`doc_toc`	Table of contents for an indexed document
`doc_get`	Fetch a specific section (~200 tokens)
`doc_query`	Keyword search across document chunks