explore-codebase

Navigate codebase structure using knowledge graph queries and semantic search.

25|4|Updated Dec 21, 2025

Search Tags:#knowledge graph #semantic search #code navigation #code architecture #call graph #codebase exploration #mcp tools

Ready to run

npx skills add https://github.com/thangchung/agent-engineering-experiment --skill explore-codebase-thangchung

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: explore-codebase
Source: https://github.com/thangchung/agent-engineering-experiment/tree/main/agentgateway-entraid-obo/.claude/skills/explore-codebase
Command: npx skills add https://github.com/thangchung/agent-engineering-experiment --skill explore-codebase-thangchung

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Understanding a large or unfamiliar codebase is slow when done by manually reading files. This Skill uses a code knowledge graph to quickly surface architecture, modules, functions, and execution flows without exhaustive manual inspection. ## Core Features & Use Cases - Architecture Discovery: Run graph statistics and architecture overview tools to see high-level community and module structure. - Semantic Code Search: Find specific functions or classes with semantic search, then trace callers, callees, and imports with graph queries. - Execution Flow Analysis: List and inspect flows to understand how execution paths move through the code. - Use Case: When onboarding to a new repository, start with graph stats and the architecture overview, then drill into the community containing the feature you need to modify, and trace its callers before making changes. ## Quick Start Use the explore-codebase skill to give me an architecture overview of this repository and trace the callers of the authentication module.

Frequently Asked Questions about explore-codebase

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I explore a large codebase quickly?▼

Start with list_graph_stats for overall metrics, then get_architecture_overview_tool for the high-level community structure. Drill into modules with list_communities_tool and get_community, and use semantic search to locate specific functions or classes.

How to find callers of a function in a codebase?▼

Use query_graph_tool with patterns like callers_of, callees_of, or imports_of to trace relationships between code entities. You can also run children_of on a file to list all its functions and classes before tracing.

What is a code knowledge graph used for?▼

A code knowledge graph models files, functions, classes, and their relationships as queryable nodes and edges. It enables semantic search, architecture overviews, and execution flow analysis without reading every source file manually.

How do I reduce token usage when exploring code with AI?▼

Always start with get_minimal_context before other graph tools and use detail_level="minimal" on all calls, escalating to standard only when needed. The target is completing a review or debug task in five or fewer tool calls and under 800 output tokens.

When should I not use graph-based code exploration?▼

Graph exploration requires the codebase to be indexed into the knowledge graph first, so it does not work on unindexed repositories. For single small files or quick one-line lookups, direct file reading may be faster.