TaskqueueSAFE
MCP tool for exposing a structured task queue to guide AI agent workflows. Great for taming an over-enthusiastic Claude.
Overview
From the repository's own README, as read at the audited commit. Badges and raw HTML are left out.
[](https://smithery.ai/server/@chriscarrollsmith/taskqueue-mcp)
MCP Task Manager (npm package: taskqueue-mcp) is a Model Context Protocol (MCP) server for AI task management. This tool helps AI assistants handle multi-step tasks in a structured way, with optional user approval checkpoints.
Features
- Task planning with multiple steps
- Progress tracking
- User approval of completed tasks
- Project completion approval
- Task details visualization
- Task status state management
- Enhanced CLI for task inspection and management
Basic Setup
Usually you will set the tool configuration in Claude Desktop, Cursor, or another MCP client as follows:
{
"tools": {
"taskqueue": {
"command": "npx",
"args": ["-y", "taskqueue-mcp"]
}
}
}To use the CLI utility, you can install the package globally and then use the following command:
npx taskqueue --help
This will show the available commands and options.
Advanced Configuration
The task manager supports multiple LLM providers for generating project plans. You can configure one or more of the following environment variables depending on which providers you want to use:
OPENAI_API_KEY: Required for using OpenAI models (e.g., GPT-4)GOOGLE_GENERATIVE_AI_API_KEY: Required for using Google's Gemini modelsDEEPSEEK_API_KEY: Required for using Deepseek models
To generate project plans using the CLI, set these environment variables in your shell:
export OPENAI_API_KEY="your-api-key" export GOOGLE_GENERATIVE_AI_API_KEY="your-api-key" export DEEPSEEK_API_KEY="your-api-key"
Or you can include them in your MCP client configuration to generate project plans with MCP tool calls:
{
"tools": {
"taskqueue": {
"command": "npx",
"args": ["-y", "taskqueue-mcp"],
"env": {
"OPEN9cb382801ef7OBSERVED · 2026-10-07Connect
Built from this server's own package name, version and transport as found in its source — not copied from anyone's documentation, so it cannot drift against a page we do not control. Replace the environment placeholders with a token scoped to the least it needs.
claude mcp add taskqueue-mcp --env DEEPSEEK_API_KEY=${DEEPSEEK_API_KEY} --env GOOGLE_GENERATIVE_AI_API_KEY=${GOOGLE_GENERATIVE_AI_API_KEY} --env OPENAI_API_KEY=${OPENAI_API_KEY} -- npx -y [email protected]{
"mcpServers": {
"taskqueue-mcp": {
"command": "npx",
"args": [
"-y",
"[email protected]"
],
"env": {
"DEEPSEEK_API_KEY": "${DEEPSEEK_API_KEY}",
"GOOGLE_GENERATIVE_AI_API_KEY": "${GOOGLE_GENERATIVE_AI_API_KEY}",
"OPENAI_API_KEY": "${OPENAI_API_KEY}"
}
}
}
}Exposed tools (14)
8 read · 4 write · 2 destructive. Blast radius: 2 tools can delete or overwrite — an agent that can be talked into calling a tool can be talked into calling this one.
| Tool | Risk | Description |
|---|---|---|
add_tasks_to_project | write | Add new tasks to an existing project. |
approve_task | read | Approve a completed task. Tasks must be marked as |
create_project | write | Create a new project with an initial prompt and a list of tasks. This is typically the first step in any workflow. |
create_task | write | Create a new task within an existing project. You can optionally include tool and rule recommendations to guide task completion. |
delete_project | destructive | Delete a project and all its associated tasks. |
delete_task | destructive | Remove a task from a project. |
finalize_project | read | Mark a project as complete. Can only be called when all tasks are both done and approved. This is typically the last step in a project workflow. |
generate_project_plan | read | Use an LLM to generate a project plan and tasks from a prompt. The LLM will analyze the prompt and any attached files to create a structured project plan. |
get_next_task | read | Get the next task to be done in a project. Returns the first non-approved task in sequence, regardless of status. The task may include toolRecommendations and ruleRecommendations fields that should be used to guide task completion. |
list_projects | read | List all projects in the system and their basic information (ID, initial prompt, task counts), optionally filtered by state (open, pending_approval, completed, all). |
list_tasks | read | List all tasks, optionally filtered by project ID and/or state (open, pending_approval, completed, all). Tasks may include tool and rule recommendations to guide their completion. |
read_project | read | Read all information for a given project, by its ID, including its tasks |
read_task | read | Get details of a specific task by its ID. The task may include toolRecommendations and ruleRecommendations fields that should be used to guide task completion. |
update_task | write | Modify a task |
Trust audit
SAFEgrade B · trust 89/100 Nothing in the source contradicts what it says it does. Grade A is reserved for packages that have also passed the behavioural sandbox.
| Layer | What it checks | Result |
|---|---|---|
| L0 | Provenance & inventory | PASS |
| L1 | Static analysis of the code | PASS |
| L2 | Instruction surface (what it tells the agent) | PASS |
| L3 | Class-specific surface | WARN |
| L4 | Behavioural (sandbox) | SKIPPED |
What the source does
- Filesystem
- declared (1 observation(s))
- Network
- none-observed
- Shell
- none-observed
- Dependencies
- not all pinned
- Secrets in source
- none-found
Findings (6)
delete_project, delete_task
import { Task, Project, TaskManagerFile } from "../../src/types/data.js";import { FileSystemService } from "../../src/server/FileSystemService.js";import { Task } from "../../../src/types/data.js";import { Task } from "../../../src/types/data.js";@ai-sdk/deepseek, @ai-sdk/google, @ai-sdk/openai, @modelcontextprotocol/sdk, ai, chalk, cli-table3, commander
Gates applied: no_behavioural_pass.
9cb382801ef7full audit observations/trust-audit/mcp-server/chriscarrollsmith__taskqueue.json · Report an issue / request a re-scanAudit history
Every audit this server has had. A grade with a past is a grade somebody is still checking.
| Date | Source | Verdict | Grade | Score | Change |
|---|---|---|---|---|---|
| 2026-10-07 | 9cb382801ef7 | SAFE | B | 89 | first audit |
Questions
What is the Taskqueue MCP server?
MCP tool for exposing a structured task queue to guide AI agent workflows. Great for taming an over-enthusiastic Claude.
What tools does Taskqueue expose?
14 in total: 8 read-only, 4 that write, and 2 that can delete or overwrite (delete_project, delete_task). Every one is listed on this page with its risk.
Is Taskqueue safe to connect to an agent?
The audit found nothing in the source that contradicts what it says it does, and graded it B (89/100). Grade A is held back for packages that have also passed a sandboxed behavioural run, which is why a clean server reads B. Separately from the audit: 2 of its tools can destroy data, so scope the token you give it to what you actually need.
What credentials does Taskqueue need?
It reads DEEPSEEK_API_KEY, GOOGLE_GENERATIVE_AI_API_KEY and OPENAI_API_KEY from the environment. Give it a token scoped to the least it needs — an agent that can be talked into calling a tool can be talked into calling it with your credentials.
How does Taskqueue run?
It speaks stdio, so it runs as a local process your client starts. It is published on npm as taskqueue-mcp at 1.4.1.
How current is this page?
The grade is for one exact copy of the source (9cb382801ef7), read on 2026-10-07. The repository is watched and re-audited when it changes.