The kill-process-mcp server allows you to manage OS processes using natural language commands through the MCP (Model Context Protocol).
Capabilities:
List Processes: View running processes with advanced filtering options (by name, user, status, CPU/memory thresholds)
Sort Processes: Order by CPU or memory usage (ascending or descending)
Kill Processes: Terminate specific processes by PID
Filtering Options: Include/exclude system processes, limit number of results
Cross-Platform: Works on macOS, Windows, and Linux
Natural Language Integration: Compatible with MCP clients like Claude Desktop for intuitive process management
Supports process management on Linux systems, allowing users to identify and terminate processes through natural language commands.
Enables listing and terminating processes running on macOS systems with filtering by CPU/memory usage, name, user, and other criteria.
Mentioned as an example use case where the MCP can be used to terminate Spotify processes by name.
Click on "Install Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@kill-process-mcp show me the top 5 processes using the most memory"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
Cross-platform MCP (Model Context Protocol) server exposing LLM-accessible tools to list and kill OS processes via natural language queries.
Perfect for shy ninjas who just want rogue processes gone: "Find and nuke the damn CPU glutton choking my system!"
Related MCP server: computer-use-windows
The following tools are exposed to MCP clients:
process_list: Lists running processes sorted by CPU or memory with optional name, user, status, CPU/memory thresholds, system-process filtering, sort order and limit
process_kill: Terminates the selected process (with extreme prejudice!)
MCP-compatible LLM client (like Claude Desktop or Cursor)
OS: macOS/Windows/Linux
Python 3.13 or higher
Libraries: mcp psutil
You can install kill-process-mcp in two ways:
Preferred: use uvx - no cloning or setup needed.
Alternative: clone the repo and set up manually.
Install uv if missing:
pip install uv
# or on macOS:
brew install uvIn case of the preferred uvx method you can now configure your MCP client (skip the cloning step below).
git clone https://github.com/misiektoja/kill-process-mcp.git
cd kill-process-mcpInstall dependencies:
uv syncRegister the kill-process-mcp as an MCP server in Claude Desktop.
Add the following to claude_desktop_config.json file if you want to use uvx method (recommended):
{
"mcpServers": {
"kill-process-mcp": {
"command": "uvx",
"args": ["kill-process-mcp@latest"]
}
}
}In case of an alternative manual method using a cloned repo:
{
"mcpServers": {
"kill-process-mcp": {
"command": "uv",
"args": [
"run",
"--directory",
"/path/to/kill-process-mcp",
"kill_process_mcp.py"
]
}
}
}Default claude_desktop_config.json location (if the file is missing - create it):
macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json
Replace /path/to/kill-process-mcp with the actual path of your project folder (remember to escape backslash characters if you're on Windows, e.g.: C:\\path\\to\\kill-process-mcp)
Restart Claude Desktop and it should be able to talk to the kill-process-mcp server.
You can check if the server is loaded by going to Profile → Settings → Connectors.
Register the kill-process-mcp as an MCP server in Cursor.
Open Cursor settings and click Tools & MCP → Add Custom MCP.
Once the mcp.json file opens, add the following if you want to use uvx method (recommended):
{
"mcpServers": {
"kill-process-mcp": {
"command": "uvx",
"args": ["kill-process-mcp@latest"]
}
}
}In case of an alternative manual method using a cloned repo:
{
"mcpServers": {
"kill-process-mcp": {
"command": "uv",
"args": [
"run",
"--directory",
"/path/to/kill-process-mcp",
"kill_process_mcp.py"
]
}
}
}Default mcp.json location:
macOS/Linux: ~/.cursor/mcp.json
Windows: %USERPROFILE%\.cursor\mcp.json
Replace /path/to/kill-process-mcp with the actual path of your project folder (remember to escape backslash characters if you're on Windows, e.g.: C:\\path\\to\\kill-process-mcp)
You should be able to talk to the kill-process-mcp server now.
You can check if the server is loaded by going to Cursor settings and clicking Tools & MCP.
If you prefer faster startup or offline use while using the uvx method, you can install a local shim once:
uv tool install kill-process-mcpThen change your LLM client config to:
{
"mcpServers": {
"kill-process-mcp": {
"command": "kill-process-mcp"
}
}
}Here are some example prompts you can use with your MCP-compatible AI assistant when interacting with this MCP server:
Kill the damn process slowing down my system!
Check my top 5 CPU parasites and flag any that look like malware
List the 3 greediest processes by RAM usage
Exterminate every process with Spotify in its name
List Alice's Python processes, max 10 entries
Which processes are over 2% CPU and 100 MB RAM
anything else your imagination brings ...
When using uvx, it automatically fetches and runs the latest published version each time your LLM client starts.
If you're using the alternative manual method with a cloned repo, update with:
cd kill-process-mcp
git pull
uv sync --reinstallWe do not pin Python. New minor versions are usually supported on day one via wheels.
If you're using the alternative manual method with a cloned repo and you hit a build error (e.g pydantic-core or rpds-py failing with a Rust toolchain message), it usually means the ecosystem is catching up with the latest Python version. In most cases this is temporary and fixed shortly by
upstream packages.
Try a clean rebuild in such case:
cd kill-process-mcp
rm -rf .venv
uv syncIf that still fails, temporarily use your previous Python minor version until compatible wheels are published (typically within a few days).
This MCP server is armed and dangerous. If you snipe the wrong process, that's on you.
Proceed with caution.
See RELEASE_NOTES.md for details.
Licensed under GPLv3. See LICENSE.
Kill the process identified by the given PID
| Name | Required | Description | Default |
|---|---|---|---|
| ctx | No | ||
| pid | Yes |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full burden of behavioral disclosure. It states the action ('kill') which implies a destructive operation, but doesn't elaborate on consequences (e.g., process termination effects, permissions required, system impact, or error handling). This is inadequate for a potentially dangerous operation.
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 a single, direct sentence with no wasted words. It's appropriately sized and front-loaded, making it easy to understand immediately.
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 destructive tool with no annotations, 0% schema coverage, and an output schema (which helps but doesn't cover behavioral aspects), the description is insufficient. It lacks critical details about safety, permissions, and usage context that are essential for responsible tool invocation.
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 description coverage is 0%, so the description must compensate. It mentions 'PID' which corresponds to the 'pid' parameter, adding some meaning. However, it doesn't explain the 'ctx' parameter at all, leaving half the parameters undocumented. The minimal compensation warrants a baseline score.
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 action ('kill') and the target ('process identified by the given PID'), which is specific and unambiguous. However, it doesn't explicitly differentiate from the sibling tool 'process_list', which would be needed for a perfect score.
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 provides no guidance on when to use this tool versus alternatives, nor any prerequisites or context for its use. It simply states what it does without indicating appropriate scenarios or warnings.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
List running processes sorted by CPU or memory with optional name, user, status, CPU/memory thresholds, system-process filtering, sort order and limit.
| Name | Required | Description | Default |
|---|---|---|---|
| ctx | No | ||
| duration | No | ||
| include_system | No | ||
| limit | No | ||
| min_cpu | No | ||
| min_memory | No | ||
| name_filter | No | ||
| sort_asc | No | ||
| sort_by | No | cpu | |
| status_filter | No | ||
| user_filter | No |
| Name | Required | Description |
|---|---|---|
| result | Yes |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
With no annotations provided, the description carries the full burden of behavioral disclosure. While it mentions the tool lists processes with sorting and filtering, it doesn't describe what the output looks like, whether this is a read-only operation, potential performance implications, or any system requirements. The description lacks behavioral context beyond the basic action.
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 a single, efficient sentence that packs substantial information about the tool's capabilities. It's appropriately sized for a tool with 11 parameters, though it could benefit from better structure by separating core purpose from parameter details.
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 complexity (11 parameters, no annotations, but with output schema), the description is moderately complete. It covers the core purpose and most parameters but lacks behavioral context and usage guidance. The presence of an output schema reduces the need to describe return values, but more context about the tool's operation would be helpful.
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?
With 0% schema description coverage for 11 parameters, the description compensates well by listing most key parameters: name, user, status, CPU/memory thresholds, system-process filtering, sort order, and limit. It provides semantic meaning for what would otherwise be completely undocumented parameters, though it doesn't cover all 11 parameters (missing 'duration' and 'ctx').
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 verb ('List') and resource ('running processes') with specific attributes ('sorted by CPU or memory'). It distinguishes from the sibling 'process_kill' by focusing on listing rather than termination. However, it doesn't explicitly contrast with the sibling tool beyond the different action.
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 mentions optional filtering capabilities but provides no guidance on when to use this tool versus alternatives. There's no mention of the sibling tool 'process_kill' or any context about when listing processes is appropriate versus terminating them. Usage is implied through parameter listing rather than explicit 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. Dates show when Glama detected each change.
v1.0.0process_killprocess_listThe two tools have clearly distinct purposes: process_kill terminates a specific process by PID, while process_list provides a comprehensive overview of running processes with filtering and sorting options. There is no overlap in functionality, making it easy for an agent to choose the correct tool based on the task.
Both tools follow a consistent verb_noun naming pattern (process_kill and process_list), using the same prefix 'process_' and clear action verbs. This uniformity makes the tool set predictable and easy to understand.
With only two tools, the server feels thin for a process management domain, as it lacks operations like process creation, monitoring, or detailed inspection. While the tools cover basic killing and listing, the scope is minimal and may limit agent capabilities in more complex scenarios.
The tool set is severely incomplete for process management. It includes kill and list operations but misses essential functions such as process creation, status updates, resource monitoring, or handling process groups. This will likely cause agent failures when tasks require more than basic termination or listing.
MCP server for generating rough-draft project plans from natural-language prompts.
GibsonAI MCP server: manage your databases with natural language
Personal assistant MCP server with search, execute, packages, jobs, secrets, and integrations.
An MCP server that integrates with Discord to provide AI-powered features.
We provide all the information about MCP servers via our MCP API.
curl -X GET 'https://glama.ai/api/mcp/v1/servers/misiektoja/kill-process-mcp'
If you have feedback or need assistance with the MCP directory API, please join our Discord server
