codingthefuturewithai/mcp-code-understanding

get_repo_file_content

Retrieve file contents or directory listings from a repository. For files, returns the complete file content. For directories, returns a non-recursive listing of immediate files and subdirectories.

PARAMETER GUIDANCE:

• repo_path: (Required) MUST match the exact format of the original input to clone_repo
  • If you cloned using a GitHub URL (e.g., 'https://github.com/username/repo'), you MUST use that identical URL here
  • If you cloned using a local directory path, you MUST use that identical local path here
  • Mismatched formats will result in 'Repository not found in cache' errors
• resource_path: (Optional) Path to the target file or directory within the repository. If not provided, it defaults to the repository's root directory.

refresh_repo

Update a previously cloned repository in MCP's cache with latest changes and trigger reanalysis. Use this to ensure analysis is based on latest code.

REQUIRED PARAMETER GUIDANCE:

• repo_path: MUST match the exact format of the original input to clone_repo
  • If you cloned using a GitHub URL (e.g., 'https://github.com/username/repo'), you MUST use that identical URL here
  • If you cloned using a local directory path, you MUST use that identical local path here
  • Mismatched formats will result in 'Repository not found in cache' errors

clone_repo

Clone a repository into the MCP server's analysis cache and initiate background analysis. Required before using other analysis endpoints like get_source_repo_map.

get_source_repo_map

Retrieve a semantic analysis map of the repository's source code structure, including file hierarchy, functions, classes, and their relationships. Repository must be previously cloned via clone_repo.

REQUIRED PARAMETER GUIDANCE:

• repo_path: MUST match the exact format of the original input to clone_repo
  • If you cloned using a GitHub URL (e.g., 'https://github.com/username/repo'), you MUST use that identical URL here
  • If you cloned using a local directory path, you MUST use that identical local path here
  • Mismatched formats will result in 'Repository not found in cache' errors

RESPONSE CHARACTERISTICS:

1. Status Types:

• "threshold_exceeded": Indicates analysis scope exceeds processing limits
• "building": Analysis in progress
• "waiting": Waiting for prerequisite operation
• "success": Analysis complete
• "error": Operation failed

2. Resource Management:

• Repository size impacts processing time and token usage
• 'max_tokens' parameter provides approximate control of response size Note: Actual token count may vary slightly above or below specified limit
• File count threshold exists to prevent overload
• Processing time scales with both file count and max_tokens Important: Clients should adjust their timeout values proportionally when:
  • Analyzing larger numbers of files
  • Specifying higher max_tokens values
  • Working with complex repositories

3. Scope Control Options:

• 'files': Analyze specific files. If only this is provided, the entire repository will be searched for matching file names.
• 'directories': Analyze all source files within specific directories.
• If BOTH 'files' and 'directories' are provided, the tool will perform an INTERSECTION, analyzing only the files named in 'files' that are also located within the specified 'directories'.

4. Response Metadata:

• Contains processing statistics and limitation details
• Provides override_guidance when thresholds are exceeded
• Reports excluded files and completion status

NOTE: This tool supports both broad and focused analysis strategies. Response handling can be adapted based on specific use case requirements and user preferences.

get_repo_structure

Retrieve directory structure and analyzable file counts for a repository to guide analysis decisions.

REQUIRED PARAMETER GUIDANCE:

• repo_path: MUST match the exact format of the original input to clone_repo
  • If you cloned using a GitHub URL (e.g., 'https://github.com/username/repo'), you MUST use that identical URL here
  • If you cloned using a local directory path, you MUST use that identical local path here
  • Mismatched formats will result in 'Repository not found in cache' errors

RESPONSE CHARACTERISTICS:

1. Directory Information:

• Lists directories containing analyzable source code
• Reports number of analyzable files per directory
• Shows directory hierarchy
• Indicates file extensions present in each location

2. Usage:

• Requires repository to be previously cloned via clone_repo
• Helps identify main code directories
• Supports planning targeted analysis
• Shows where analyzable code is located

NOTE: Use this tool to understand repository structure and choose which directories to analyze in detail.

get_repo_critical_files

Identify and analyze the most structurally significant files in a repository to guide code understanding efforts.

REQUIRED PARAMETER GUIDANCE:

• repo_path: MUST match the exact format of the original input to clone_repo
  • If you cloned using a GitHub URL (e.g., 'https://github.com/username/repo'), you MUST use that identical URL here
  • If you cloned using a local directory path, you MUST use that identical local path here
  • Mismatched formats will result in 'Repository not found in cache' errors

RESPONSE CHARACTERISTICS:

1. Analysis Metrics:

   • Calculates importance scores based on:
     • Function count (weight: 2.0)
     • Total cyclomatic complexity (weight: 1.5)
     • Maximum cyclomatic complexity (weight: 1.2)
     • Lines of code (weight: 0.05)
   • Provides detailed metrics per file
   • Ranks files by composite importance score

2. Resource Management:

   • Repository must be previously cloned via clone_repo
   • Analysis performed on-demand using Lizard
   • Efficient for both small and large codebases
   • Supports both full-repo and targeted analysis

3. Scope Control Options:

   • 'files': Analyze specific files. If only this is provided, the entire repository will be searched for matching file names.
   • 'directories': Analyze all source files within specific directories.
   • If BOTH 'files' and 'directories' are provided, the tool will perform an INTERSECTION, analyzing only the files named in 'files' that are also located within the specified 'directories'.
   • 'limit': Control maximum results returned.

4. Response Metadata:

   • Total files analyzed
   • Analysis completion status

NOTE: This tool is designed to guide initial codebase exploration by identifying structurally significant files. Results can be used to target subsequent get_source_repo_map calls for detailed analysis.

get_repo_documentation

Retrieve and analyze documentation files from a repository, including README files, API docs, design documents, and other documentation. Repository must be previously cloned via clone_repo.

REQUIRED PARAMETER GUIDANCE:

• repo_path: MUST match the exact format of the original input to clone_repo
  • If you cloned using a GitHub URL (e.g., 'https://github.com/username/repo'), you MUST use that identical URL here
  • If you cloned using a local directory path, you MUST use that identical local path here
  • Mismatched formats will result in 'Repository not found in cache' errors