Analyze Codebase

Overview

Use the atomicviz MCP server to analyze a codebase for issues, dependency structure, and function-call flow. This skill focuses on concrete, tool-backed findings and visualizable graphs.

Example Prompts

• •What is not hooked up right in the codebase?
• •Is there placeholder code somewhere?
• •Analyze the codebase and map out function calls.
• •Trace the impact of changing FooService in src/foo.ts.
• •Map all database usage and find naming conflicts.

Quick Start

1. •Ask for scope if unclear (repo root or specific folders/files).
2. •Run issue detection and structure scans:
   • •detect_issues on the target path.
   • •analyze_dependencies or show_architecture for structure.
3. •For call flow:
   • •visualize_function_calls for a file/function.
   • •trace_impact for change ripple.
4. •For deeper audits:
   • •comprehensive_file_map, find_naming_conflicts, map_database_usage.

Tool Selection Guide

Issues / placeholders / dead code

• •Use detect_issues with issueTypes: ['unused','dead-code','complexity','coupling'].
• •Follow with comprehensive_file_map when you need global context.

Function calls / control flow

• •Use visualize_function_calls for a file or function.
• •Use trace_impact to see all affected call sites.

Architecture / dependencies

• •Use analyze_dependencies for module relationships.
• •Use show_architecture or simplify_graph for a high-level view.

Data + database usage

• •Use map_database_usage to map table/column usage.

Naming + structure consistency

• •Use find_naming_conflicts and compare_structures (before/after).

Large repos

• •Use analyze_large_project to chunk analysis.

Output Expectations

• •Provide a short findings list with file paths and why they matter.
• •When graphs are produced, summarize the key nodes/edges and risk areas.
• •If you used multiple tools, explain how results corroborate each other.

Notes

• •Many tools require absolute paths; normalize inputs.
• •Keep scopes narrow for faster, clearer results.