MCP-сервер для инспекции, управления и завершения локальных серверов разработки. Перестаньте гадать, что запущено на :3000.
Работает в паре с terminal-history-mcp — вместе они дают вашему ИИ-агенту полную память о вашей среде разработки: что вы запускали и что запущено сейчас.
Каждый разработчик сталкивается с этим ежедневно:
Error: listen EADDRINUSE :::3000 — что занимает порт?
5 забытых PID node / vite / next с прошлой недели, потребляющих оперативную память
Переключение между проектами → нет понимания, какие серверы разработки все еще запущены
lsof -i :3000, kill -9 <pid>, повторить
localhost-mcp позволяет сделать это одним вызовом инструмента.
Related MCP server: Terminal History MCP
npm install -g localhost-mcpПодключите к Claude Code:
claude mcp add --scope user localhost -- localhost-mcpИли к любому MCP-совместимому клиенту. Команда запускается как stdio MCP-сервер.
Инструмент | Назначение |
| Все активные серверы разработки с портом, pid, фреймворком, проектом, временем работы, памятью, CPU |
| Инспекция конкретного порта — кто его занимает |
| Завершение по pid или порту. По умолчанию — предварительный просмотр; передайте |
| Обнаружение долго работающих, простаивающих или ресурсоемких серверов разработки |
| Почему порт X занят + 5 свободных альтернатив рядом |
{
"port": 3000,
"pid": 48211,
"process": "node",
"cmdline": "next dev",
"cwd": "/Users/me/code/myapp",
"project_name": "myapp",
"framework": "next.js",
"uptime_seconds": 14523,
"memory_mb": 412,
"cpu_pct": 0.3,
"user": "me"
}kill_server по умолчанию работает в режиме предварительного просмотра. Необходимо передать confirm=true.
Отказ от завершения PID < 1000 (системные процессы).
Отказ от завершения процессов вне белого списка разработки (node, python, ruby, go, deno, bun, php, java, rails, vite, next и т.д.).
Сначала отправляется SIGTERM, через 5 секунд ожидания — SIGKILL.
next.js, vite, nuxt, remix, astro, webpack-dev-server, esbuild, create-react-app, express, fastify, koa, hono, rails, django, flask, fastapi, uvicorn, gunicorn, deno, bun, php-builtin, jekyll, hugo.
При использовании общей командной строки (node server.js) выполняется поиск по package.json.
macOS — полная поддержка (использует lsof)
Linux — полная поддержка (использует lsof + /proc)
Windows — только базовое сканирование портов (использует netstat); определение рабочей директории / фреймворка ограничено
localhost-mcp list # JSON list of all dev servers
localhost-mcp zombies # JSON list of zombie candidates
localhost-mcp # Start MCP stdio servergit clone https://github.com/hasanjahidul/localhost-mcp.git
cd localhost-mcp
npm install
npm run build
node dist/cli.js listMIT
Read-only. Flags dev servers that look abandoned — all three of: uptime > 6h AND CPU < 1% AND memory > 100MB. By default it excludes known always-on noise (VS Code server, language servers, other MCP servers, postgres/redis, sidekiq, etc.); set include_excluded: true to also list those. Each candidate comes with the reasons it matched. This tool never kills anything — pass results to kill_server. Returns { count, candidates[] }.
| Name | Required | Description | Default |
|---|---|---|---|
| include_excluded | No | Also include IDE/LSP/agent/DB processes that match the heuristic but are normally filtered out. Default false. |
| Name | Required | Description |
|---|---|---|
| count | No | |
| candidates | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
The description complements the annotations (readOnlyHint, destructiveHint, idempotentHint) by explicitly stating 'Read-only' and 'This tool never kills anything'. It details the exact heuristic logic and default filtering behavior, adding substantial context beyond what annotations provide.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is highly concise, with three sentences packed with essential information. It front-loads core purpose, then covers behavior, parameter, safety, and output format without unnecessary words.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the tool's simplicity (single boolean parameter, clear heuristic), the description fully covers matching criteria, default exclusion behavior, the parameter's effect, relationship to sibling tools, and the return format. No gaps remain.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage for the single parameter (include_excluded) is 100%, and the schema already includes a thorough description. The tool description restates the parameter's effect but does not add significant new semantics beyond what the schema provides.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states that the tool flags dev servers that appear abandoned, and precisely defines the three matching criteria (uptime > 6h, CPU < 1%, memory > 100MB). It distinguishes itself from sibling tools by explicitly stating it never kills anything and that results should be passed to kill_server.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explains that the tool is read-only and should be used to identify candidates, while kill_server is for termination. It also describes the default exclusion of noise processes and the include_excluded parameter, giving clear guidance on when to use this tool versus alternatives.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Terminates a local dev server by pid or port. DESTRUCTIVE. Safe by default: with no confirm, this is a DRY RUN — it reports what it would kill and changes nothing. Pass confirm: true to actually terminate: sends SIGTERM, waits up to 5s, then escalates to SIGKILL (or SIGKILL immediately if force: true). Refuses PIDs below 1000 and processes that don't look like dev servers. Provide exactly one of pid or port. Returns what was (or would be) killed and the signal used.
| Name | Required | Description | Default |
|---|---|---|---|
| pid | No | Process ID to kill. Mutually exclusive with `port`. | |
| port | No | TCP port whose listening process should be killed. Mutually exclusive with `pid`. | |
| force | No | Skip SIGTERM and send SIGKILL immediately. Default false. | |
| confirm | No | Must be true to actually terminate the process. When false/omitted the call is a dry run. |
| Name | Required | Description |
|---|---|---|
| pid | No | |
| port | No | |
| killed | No | |
| signal | No | `SIGTERM`, `SIGKILL`, or null on a dry run. |
| dry_run | No | |
| message | No | |
| process | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Rich behavioral details beyond annotations: dry-run behavior, escalation from SIGTERM to SIGKILL with timeout, force flag effect, safety guardrails (refuses low PIDs and non-dev-server processes). Annotations provide destructiveHint, but description adds deep context without contradiction.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two dense, front-loaded sentences with zero redundancy. Every clause adds essential information (alternatives, safety, escalation, constraints).
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the presence of output schema (not shown) and annotations, the description covers behavior, constraints, safety mechanisms, and parameter relationships completely. No gaps for an AI agent to misunderstand.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema already describes all 4 parameters fully (100% coverage). The description adds value by contextualizing the mutex between pid/port, explaining the safe default behavior, and clarifying the effect of confirm vs force beyond the schema descriptions.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description uses active verbs ('terminates', 'reports', 'sends') and clearly specifies the resource ('local dev server') and scope ('by pid or port'). It distinguishes from siblings like list_dev_servers and find_zombies by focusing solely on termination.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Explicitly states 'Provide exactly one of pid or port' and explains the dry-run default and confirm requirement. While it doesn't contrast with siblings (not needed as they serve different purposes), it gives clear context for when to use this tool.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Read-only. Lists every local development server (next, vite, nuxt, remix, astro, rails, django, flask, express, deno, bun, etc.) currently LISTENING on a TCP port. For each: port, pid, process name, command line, working directory, project name, detected framework, uptime, memory (MB), CPU %, and owning user. Uses lsof on macOS/Linux and netstat on Windows; on Windows the cwd/framework fields are limited. Takes no arguments. Returns { count, servers[] }.
| Name | Required | Description | Default |
|---|---|---|---|
No parameters | |||
| Name | Required | Description |
|---|---|---|
| count | No | |
| servers | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
The description adds significant behavioral context beyond annotations: it is read-only, uses lsof/netstat based on OS, notes Windows limitations on cwd/framework fields, and details the full output schema. Annotations confirm read-only, and no contradictions exist.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is three sentences with no fluff. It starts with the purpose, lists output fields, and ends with platform details and return structure. Every sentence earns its place.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a no-parameter tool, the description is exhaustive. It covers platform differences, limitations, output format, and all fields. The output schema exists and is described. No gaps remain.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
There are no parameters; the description correctly states 'takes no arguments'. With 0 parameters, baseline is 4, and the description adds no extra param information but confirms the schema.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool lists local development servers listening on TCP ports, with a specific verb 'lists' and resource 'dev servers'. It includes a long list of frameworks and distinguishes itself from siblings like find_zombies or kill_server by focusing on active listening servers.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies usage when one wants to enumerate dev servers, but does not explicitly state when to use it over alternatives like port_info or port_conflict. No exclusion criteria or alternative tool names are mentioned.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Read-only. Diagnoses an EADDRINUSE situation: given a port, returns the dev server currently blocking it plus 5 free alternative ports nearby. Use it when a listen EADDRINUSE error fires and you want both the culprit and a port to switch to.
| Name | Required | Description | Default |
|---|---|---|---|
| port | Yes | The contended TCP port, e.g. 3000. |
| Name | Required | Description |
|---|---|---|
| port | No | |
| blocked_by | No | The dev-server record holding the port (see list_dev_servers), or null if actually free. |
| alternatives | No | Up to 5 nearby free ports. |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint, destructiveHint false, and idempotentHint true. The description adds that it returns the blocking server and free ports, which is useful context beyond the annotations. No contradictions.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences: the first declares read-only behavior and functionality, the second provides usage guidance. No extraneous words; each sentence earns its place.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
For a simple tool with one parameter and an output schema, the description fully covers purpose, trigger condition, and return value. No gaps remain.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100% for the single parameter. The description repeats the schema's description ('The contended TCP port, e.g. 3000.') without adding new meaning. Baseline 3 is appropriate since the schema already provides sufficient detail.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool diagnoses an EADDRINUSE situation by returning the blocking dev server and 5 free alternative ports. Its verb 'diagnose' and resource 'port conflict' are specific, and it distinguishes itself from sibling tools (e.g., kill_server, find_zombies) that serve different purposes.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explicitly advises using this tool when a 'listen EADDRINUSE' error occurs, specifying the desired outcome (culprit and free ports). While it doesn't list alternatives or when not to use, the context is clear enough for an agent to decide.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Read-only. Inspects a single TCP port and reports the dev server holding it (same fields as list_dev_servers). If nothing is listening, or the listener is not a recognised dev process, returns { port, status: "free" }. Useful for answering "what's on :3000?" before starting or killing something.
| Name | Required | Description | Default |
|---|---|---|---|
| port | Yes | TCP port number, 1–65535, e.g. 3000. |
| Name | Required | Description |
|---|---|---|
| port | No | |
| status | No |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint=true, destructiveHint=false, idempotentHint=true, and openWorldHint=true. The description adds value by explaining the return format (same fields as `list_dev_servers`) and handling edge cases (nothing listening or not a recognised dev process). No contradiction with annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is three sentences, with the key verb 'Read-only' front-loaded. Every sentence adds necessary information: purpose, return format details, and usage scenario. No wasted words.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Given the presence of annotations (read-only, idempotent, non-destructive) and an output schema (implied), the description is complete. It covers behavior for all inputs (port in range) and edge cases (free port, unrecognised process). The tool's scope is well-defined.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100% with a well-described parameter (port). The description adds 'TCP port number, 1–65535, e.g. 3000,' which mirrors the schema. Since the schema already does the heavy lifting, the description does not add significant new semantics beyond clarifying the example.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description clearly states the tool inspects a single TCP port and reports the dev server holding it. It distinguishes from sibling tools like `list_dev_servers` (which lists all servers) and implies a specific function ('what's on :3000?').
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explicitly says 'Useful for answering "what's on :3000?" before starting or killing something,' which provides clear context for when to use. However, it does not explicitly mention when not to use or list alternatives, though sibling tool names provide some guidance.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v0.1.1find_zombies2 fields changedInput schema / properties / include_excluded / descriptionPrevious value: -"Include IDE/LSP/agent/DB processes that match the zombie heuristic. Default false."New value: +"Also include IDE/LSP/agent/DB processes that match the heuristic but are normally filtered out. Default false."Output schema / (root)Previous value: -nullNew value: +{
+ "properties": {
+ "candidates": {
+ "items": {
+ "properties": {
+ "cmdline": {
+ "description": "Full command line, e.g. `next dev`.",
+ "type": "string"
+ },
+ "cpu_pct": {
+ "description": "Instantaneous CPU percent from `ps`.",
+ "type": "number"
+ },
+ "cwd": {
+ "description": "Working directory of the process (best-effort; may be empty on Windows).",
+ "type": "string"
+ },
+ "excluded": {
+ "description": "True if it only appears because `include_excluded` was set.",
+ "type": "boolean"
+ },
+ "framework": {
+ "description": "Detected framework, e.g. `next.js`, `vite`, `rails`; `unknown` if not recognised.",
+ "type": "string"
+ },
+ "memory_mb": {
+ "description": "Resident set size in MB.",
+ "type": "number"
+ },
+ "pid": {
+ "type": "number"
+ },
+ "port": {
+ "type": "number"
+ },
+ "process": {
+ "description": "Process name, e.g. `node`, `python`.",
+ "type": "string"
+ },
+ "project_name": {
+ "description": "Basename of `cwd`, or the npm package name when detected.",
+ "type": "string"
+ },
+ "reasons": {
+ "description": "Why this process was flagged, e.g. `uptime 14h`, `cpu 0.1%`, `mem 412MB`.",
+ "items": {
+ "type": "string"
+ },
+ "type": "array"
+ },
+ "uptime_seconds": {
+ "type": "number"
+ },
+ "user": {
+ "type": "string"
+ }
+ },
+ "type": "object"
+ },
+ "type": "array"
+ },
+ "count": {
+ "type": "number"
+ }
+ },
+ "type": "object"
+}kill_server5 fields changedInput schema / properties / confirm / descriptionPrevious value: -"Must be true to actually kill"New value: +"Must be true to actually terminate the process. When false/omitted the call is a dry run."Input schema / properties / force / descriptionPrevious value: -"Use SIGKILL immediately, skip SIGTERM"New value: +"Skip SIGTERM and send SIGKILL immediately. Default false."Input schema / properties / pid / descriptionPrevious value: -"Process ID"New value: +"Process ID to kill. Mutually exclusive with `port`."Input schema / properties / port / descriptionPrevious value: -"TCP port (alternative to pid)"New value: +"TCP port whose listening process should be killed. Mutually exclusive with `pid`."Output schema / (root)Previous value: -nullNew value: +{
+ "properties": {
+ "dry_run": {
+ "type": "boolean"
+ },
+ "killed": {
+ "type": "boolean"
+ },
+ "message": {
+ "type": "string"
+ },
+ "pid": {
+ "type": "number"
+ },
+ "port": {
+ "type": "number"
+ },
+ "process": {
+ "type": "string"
+ },
+ "signal": {
+ "description": "`SIGTERM`, `SIGKILL`, or null on a dry run.",
+ "type": "string"
+ }
+ },
+ "type": "object"
+}list_dev_servers1 field changedOutput schema / (root)Previous value: -nullNew value: +{
+ "properties": {
+ "count": {
+ "type": "number"
+ },
+ "servers": {
+ "items": {
+ "properties": {
+ "cmdline": {
+ "description": "Full command line, e.g. `next dev`.",
+ "type": "string"
+ },
+ "cpu_pct": {
+ "description": "Instantaneous CPU percent from `ps`.",
+ "type": "number"
+ },
+ "cwd": {
+ "description": "Working directory of the process (best-effort; may be empty on Windows).",
+ "type": "string"
+ },
+ "framework": {
+ "description": "Detected framework, e.g. `next.js`, `vite`, `rails`; `unknown` if not recognised.",
+ "type": "string"
+ },
+ "memory_mb": {
+ "description": "Resident set size in MB.",
+ "type": "number"
+ },
+ "pid": {
+ "type": "number"
+ },
+ "port": {
+ "type": "number"
+ },
+ "process": {
+ "description": "Process name, e.g. `node`, `python`.",
+ "type": "string"
+ },
+ "project_name": {
+ "description": "Basename of `cwd`, or the npm package name when detected.",
+ "type": "string"
+ },
+ "uptime_seconds": {
+ "type": "number"
+ },
+ "user": {
+ "type": "string"
+ }
+ },
+ "type": "object"
+ },
+ "type": "array"
+ }
+ },
+ "type": "object"
+}port_conflict2 fields changedInput schema / properties / port / descriptionAdded value: +"The contended TCP port, e.g. 3000."Output schema / (root)Previous value: -nullNew value: +{
+ "properties": {
+ "alternatives": {
+ "description": "Up to 5 nearby free ports.",
+ "items": {
+ "type": "number"
+ },
+ "type": "array"
+ },
+ "blocked_by": {
+ "description": "The dev-server record holding the port (see list_dev_servers), or null if actually free.",
+ "type": [
+ "object",
+ "null"
+ ]
+ },
+ "port": {
+ "type": "number"
+ }
+ },
+ "type": "object"
+}port_info2 fields changedInput schema / properties / port / descriptionPrevious value: -"TCP port number"New value: +"TCP port number, 1–65535, e.g. 3000."Output schema / (root)Previous value: -nullNew value: +{
+ "description": "Either a dev-server record (see list_dev_servers) or `{ port, status: \"free\" }`.",
+ "properties": {
+ "port": {
+ "type": "number"
+ },
+ "status": {
+ "type": "string"
+ }
+ },
+ "type": "object"
+}v0.1.0find_zombieskill_serverlist_dev_serversport_conflictport_infoScored across 5 tools
Each tool has a clearly distinct purpose: listing servers, inspecting a single port, finding zombies, killing servers, and diagnosing port conflicts. There is no functional overlap.
Tool names use snake_case, and most follow a verb_noun pattern (find_zombies, kill_server, list_dev_servers). However, port_conflict and port_info are noun-first, which is a minor inconsistency.
With 5 tools, the set is well-scoped for managing local dev servers: query, inspect, cleanup, and conflict resolution. No tool feels superfluous or missing.
The tools cover listing, inspecting, zombie identification, killing, and conflict diagnosis. There is a minor gap in that bulk operations (e.g., kill all zombies) are not directly supported, but the intended workflow is clear.
2,000+ MCP servers read at source level. Know what one does before you connect. Free, no key.
Scan any MCP server for tool-poisoning, security, auth & license. Trust score before install.
Self-hosted MCP server: 26 deterministic dev, security, and EVM tools.
A MCP server built for developers enabling Git based project management with project and personal…