jcodemunch via mcporter

Query jcodemunch-mcp indexed codebases using bunx mcporter call and bunx mcporter list. This avoids loading the full MCP server into the agent's context while still accessing all of jcodemunch's code intelligence.

Prerequisites

jcodemunch-mcp must be running (either as a configured MCP server or reachable via stdio). Repos must be indexed before querying — use index_folder or index_repo to index first.

How to call

All calls go through mcporter's stdio transport:

# List available tools
bunx mcporter list --stdio "uvx jcodemunch-mcp"

# Call a tool (function-call syntax)
bunx mcporter call --stdio "uvx jcodemunch-mcp" 'list_repos()'

# Call with arguments
bunx mcporter call --stdio "uvx jcodemunch-mcp" 'search_symbols(repo: "jsed", query: "parse")'

# JSON output for structured processing
bunx mcporter call --stdio "uvx jcodemunch-mcp" 'list_repos()' --output json

The --stdio "uvx jcodemunch-mcp" flag tells mcporter to launch jcodemunch as a stdio subprocess.

Tool reference

Orientation (start here on an unfamiliar repo)

Tool	Purpose	Example
list_repos()	List all indexed repos	'list_repos()'
suggest_queries(repo)	Entry-point files, top keywords, example queries	'suggest_queries(repo: "jsed")'
get_repo_outline(repo)	High-level: dirs, file counts, languages, symbol counts	'get_repo_outline(repo: "jsed")'
get_file_tree(repo)	File tree, optionally filtered by path prefix	'get_file_tree(repo: "jsed", path_prefix: "src/utils")'

Symbol search & inspection

Tool	Purpose	Example
search_symbols(repo, query)	Search across repo by name, signature, summary	'search_symbols(repo: "jsed", query: "parse", max_results: 5)'
get_file_outline(repo, file_path)	All symbols in a file with signatures	'get_file_outline(repo: "jsed", file_path: "src/main.ts")'
get_symbol_source(repo, symbol_id)	Full source of a symbol	'get_symbol_source(repo: "jsed", symbol_id: "src/main.ts::parseToken")'
get_context_bundle(repo, symbol_id)	Source + imports, deduped. Use output_format: "markdown" for paste-ready	'get_context_bundle(repo: "jsed", symbol_id: "src/main.ts::parseToken", output_format: "markdown")'
get_related_symbols(repo, symbol_id)	Heuristic clustering: co-location, shared importers, name overlap	'get_related_symbols(repo: "jsed", symbol_id: "src/main.ts::parseToken")'

Text search

Tool	Purpose	Example
search_text(repo, query)	Full-text search across file contents, supports regex	'search_text(repo: "jsed", query: "TODO", context_lines: 2)'

Dependency & impact analysis

Tool	Purpose	Example
get_dependency_graph(repo, file)	File-level import graph, up to 3 hops	'get_dependency_graph(repo: "jsed", file: "src/main.ts", direction: "both")'
find_importers(repo, file_path)	What files import a given file	'find_importers(repo: "jsed", file_path: "src/utils/helpers.ts")'
find_references(repo, identifier)	Where is an identifier used	'find_references(repo: "jsed", identifier: "parseToken")'
check_references(repo, identifier)	Is an identifier referenced anywhere (dead-code detection)	'check_references(repo: "jsed", identifier: "oldFunction")'
get_blast_radius(repo, symbol)	All files affected by changing a symbol	'get_blast_radius(repo: "jsed", symbol: "parseToken")'
get_class_hierarchy(repo, class_name)	Inheritance tree (ancestors + descendants)	'get_class_hierarchy(repo: "jsed", class_name: "BaseParser")'

Indexing

Tool	Purpose	Example
index_folder(path)	Index a local folder	'index_folder(path: "/Users/danb/projects/myrepo")'
index_repo(url)	Index a GitHub repo	'index_repo(url: "owner/repo")'
index_file(path)	Re-index a single file (faster than full re-index)	'index_file(path: "/Users/danb/projects/myrepo/src/changed.ts")'

Diffing

Tool	Purpose	Example
get_symbol_diff(repo_a, repo_b)	Diff symbol sets between two indexed snapshots	'get_symbol_diff(repo_a: "jsed-main", repo_b: "jsed-feature")'

search_symbols detail levels

The detail_level parameter controls result verbosity:

• compact — id/name/kind/file/line only (~15 tokens each). Best for broad discovery.
• standard (default) — includes signatures and summaries.
• full — inlines source code and docstrings. Equivalent to search + get_symbol_source in one call.

Use token_budget instead of max_results when you want to pack as many results as possible into a fixed context window.

Workflow patterns

"What is this repo?"

1. list_repos() — find the repo identifier
2. suggest_queries(repo) — get entry points and example queries
3. get_repo_outline(repo) — directory structure and language breakdown

"Where is X used?"

1. find_references(repo, identifier: "X") — import-level references
2. search_text(repo, query: "X") — full-text occurrences (catches string refs, comments)

"What would break if I change X?"

1. get_blast_radius(repo, symbol: "X") — confirmed + potential affected files
2. get_dependency_graph(repo, file: "...", direction: "importers") — who depends on this file

"Show me everything about function X"

1. get_context_bundle(repo, symbol_id: "...", output_format: "markdown") — source + imports, paste-ready

Repo identifiers

Repos are identified by their display name (e.g., "jsed", "@oneput") or their full identifier (e.g., "local/jsed-2fd72aec"). Use list_repos() to discover available identifiers.

Batch operations

Several tools support batch mode for efficiency:

• get_file_outline — pass file_paths: [...] instead of file_path
• get_symbol_source — pass symbol_ids: [...] instead of symbol_id
• find_importers — pass file_paths: [...] instead of file_path
• find_references — pass identifiers: [...] instead of identifier
• check_references — pass identifiers: [...] instead of identifier