MCP SERVER · OFFICIAL PROJECTRead one function instead of a 700-line file — symbol-level code retrieval, blast-radius analysis and edit-safety checks over an indexed repository.
LABSCO SUMMARY
That gap is what this closes — a symbol instead of a file, an outline instead of contents. It helps least where an edit genuinely needs the whole file, and most on large or unfamiliar repositories, which is also where the blast-radius and edit-safety answers have no equivalent in a plain read-and-grep loop.
A code-intelligence layer over a locally indexed repository. The README counts 90-plus tools; a fresh connection lists six, among them a three-tool front door — menu, order, route — that reaches the rest on demand. Instead of reading whole files, an agent asks for the symbol, the outline, or the structural answer it needs. Indexes live under ~/.code-index/, and 70-plus languages are parsed through tree-sitter, monorepos included.
get_symbol_source, get_file_outline, get_file_content, search_symbols)assemble_task_context, plan_turn, get_ranked_context)find_importers, get_blast_radius, get_call_hierarchy, find_dead_code, get_changed_symbols, get_hotspots, search_ast){old_text, new_text} blocks from the refactoring planner (check_edit_safe, check_delete_safe, get_pr_risk_profile, plan_refactoring, check_references)stop_rule.terminal flag on the two safety checks: true means no further call changes the verdict, so re-running the reference checks to be sure is wasted work — it means final, not safe. False names the specific thing that would change the answerget_session_stats)defer_loading set once for the whole serveruv tool install jcodemunch-mcp then jcodemunch-mcp init, which detects Claude Code, Claude Desktop, Cursor, Windsurf and Continue, writes their config entries, installs the prompt policy so the agent actually uses it, and optionally indexes the project. uvx jcodemunch-mcp runs it with nothing left on disk — but the optional enforcement hooks resolve the executable by name from a minimal-PATH subshell, so those need a real install. It is local-first: indexes stay under ~/.code-index/, and the only default network behaviour is an anonymous savings counter carrying a random id and aggregate token counts, turned off with share_savings: false. The tool-schema deferral through Anthropic's connector needs the server over sse or streamable-http, not the default stdio setup.
One command — uv tool install jcodemunch-mcp && jcodemunch-mcp init
USE CASES
Find the symbol, then read just its body. Thirty lines instead of seven hundred, and the agent still sees the exact bytes it is about to change.
Ask for the blast radius and the importers rather than grepping for the name and hoping. It is the question a plain read-and-grep loop cannot answer at all.
Run the safety check and read the stop rule. Terminal means the verdict will not move, so stop checking — false names the one thing that would change it, which is where to look next.
Assemble the task's context in one call and let it classify the intent and pull the anchor symbols, rather than opening files until the shape appears.
Get the plan back as edit-ready old-text and new-text blocks, with the risk profile of the resulting change alongside it.
Ask for dead code and for the hotspots the churn is concentrated in. Both are structural answers, and both are tedious to reach by reading.
Collapse the surface to a three-tool front door with the server-side counter, or defer the schemas host-side through tool search if your client supports it. Pick one — you do not need both.
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
announce_modelAgent self-reports its active model identifier. Server resolves to a tier via model_tier_map (fuzzy: normalize → exact → glob → substring → '*' → 'full') and narrows the exposed tool list accordingly. Idempotent: a second call with the same model is a cheap no-op. Prefer calling plan_turn(model=...) for routine per-task use; use announce_model as a fallback when plan_turn is not appropriate for the current task.
jcodemunch_guideReturn the version-current CLAUDE.md / AGENT.md policy snippet for jcodemunch-mcp — the same text produced by jcodemunch-mcp claude-md --generate. Lets an agent keep a one-line CLAUDE.md (e.g. "Call jcodemunch_guide and strictly follow its instructions.") instead of pasting a static snippet that drifts from the installed version. Idempotent, no repo context required. Matches the active tool surface, tier and disabled_tools — list 'jcodemunch_guide' in disabled_tools to hide it.
menuDiscover catalog actions without keeping the full tool catalog resident: menu(query?). Returns compact rows (action, summary, required args, state_changing). With no query, lists the catalog. Pair with 'order' to dispatch the chosen action.
orderDispatch any jcodemunch action by name: order(action, args). The single-verb front door to the full tool catalog. Read-only by default — actions that change index/session state require allow_state_change=true, and execution/file-write verbs are refused. For exploration questions ('how does X work'), order('get_ranked_context', {repo, query, token_budget}) answers in ONE call — prefer it over chained search/outline/source hops; add compress=true to fit more symbols in the same budget. Call 'menu' to discover actions, or 'route' to pick one from a task.
routeMap a natural-language task to the best catalog action(s): route(task, repo?, execute?). Returns ranked recommendations with ready-to-run argument templates. With execute=true, dispatches the top recommendation and returns its result in the same call, collapsing discover-then-call into one round-trip. Recommends assemble_task_context / plan_turn for context-gathering intents.
set_tool_tierExplicit tier override for the current session. Narrows or widens the exposed tool list to 'core' / 'standard' / 'full'. Prefer plan_turn(model=...) for routine per-task use; use set_tool_tier only when you need an explicit override (e.g. escalate mid-task to 'full' after a capability-gated failure).
BEFORE YOU INSTALL
6 tools when we connected.
uvx on PATH
All 4 documented clients run the server on your machine.
CONTEXT COST
Rated by the size of the 6 tool definitions that ride along on every turn, not by how many there are
INSTALLATION
uv tool install jcodemunch-mcp
jcodemunch-mcp initclaude mcp add -s user jcodemunch -- uvx jcodemunch-mcp{
"servers": {
"jcodemunch": {
"command": "uvx",
"args": [
"jcodemunch-mcp"
]
}
}
}{
"mcpServers": {
"jcodemunch-mcp": {
"command": "uvx",
"args": [
"jcodemunch-mcp"
]
}
}
}MCP servers
n24q02m
rohitdas-ai
NitroRCr
cq27-dev