go-navigator-read
Use Go Navigator MCP to explore Go code semantically (packages, symbols, definitions/references, context, metrics, deps) and return file/line/snippets; do not modify code.
When & Why to Use This Skill
The Go Navigator (READ) skill is a high-performance Model Context Protocol (MCP) tool designed for deep semantic exploration of Go (Golang) codebases. It enables AI agents to move beyond simple text searches by providing structured access to symbols, definitions, references, and dependency graphs. By offering precise file locations, code snippets, and complexity metrics, it provides the essential context needed for comprehensive code analysis and architectural understanding without altering the source code.
Use Cases
- Codebase Onboarding: Quickly map out the structure of unfamiliar Go repositories by listing packages, project schemas, and entry points.
- Impact Analysis: Trace symbol references and dependency graphs to evaluate how changes to a specific function or type might affect the broader system.
- Interface Discovery: Effortlessly identify all concrete implementations of a specific Go interface across the entire workspace to understand polymorphic logic.
- Technical Debt Auditing: Generate complexity summaries and dead code reports to pinpoint high-risk areas or unused components requiring refactoring.
| name | go-navigator-read |
|---|---|
| description | Use Go Navigator MCP to explore Go code semantically (packages, symbols, definitions/references, context, metrics, deps) and return file/line/snippets; do not modify code. |
Go Navigator (READ) — semantic navigation only
Scope rules
- READ-ONLY: never call renameSymbol or rewriteAst (or any tool that changes files).
- Prefer semantic tools over raw text search.
Defaults
- Use "dir": "." unless the user specifies another root.
- For any "package" field, use the exact import path returned by listPackages (go list style).
Recommended workflows
Orientation in a new repo
- listPackages { "dir": "." }
- getProjectSchema { "dir": ".", "depth": "standard" } (optional)
- getMetricsSummary { "dir": ".", "package": "" } (optional)
- getDependencyGraph { "dir": ".", "package": "" } (optional)
Understand a symbol
- getDefinitions { "dir": ".", "ident": "" }
- getReferences { "dir": ".", "ident": "" }
- getSymbolContext { "dir": ".", "ident": "", "kind": "<func|type|...>" } (preferred)
- getFunctionSource / getStructInfo / getFileInfo for deeper inspection
Interfaces & implementations
- listInterfaces { "dir": ".", "package": "" }
- getImplementations { "dir": ".", "name": "" }
Complexity, dead code, summary
- getComplexityReport { "dir": ".", "package": "" }
- getDeadCodeReport { "dir": ".", "package": "", "limit": 10 }
- getMetricsSummary { "dir": ".", "package": "" }
Large result sets
- Use limit/offset pagination where supported (getDefinitions/getReferences).
- Return the most relevant files first.
Output format
- Always include file paths + line numbers.
- Include minimal snippets (1–5 lines) unless asked for full bodies.
- Summarize findings and propose next tool calls (still read-only).