comment_job
Comment on a job by ID of job
LABSCO SUMMARY
Both halves of Rework — board tasks and workflow jobs — with the field definitions readable before you create anything.
The workflow tools give you the stage list and the field schema for a workflow, including which fields are required and what options they accept, so a job can be created correctly on the first attempt rather than rejected. Name-to-id resolution is a separate step by design: look the person up, take the username and id, then pass them into the create or filter call.
What it is
A server over the Rework platform covering both halves of it: project boards with tasks, and workflows with jobs that move through stages.
What you get
- Tasks created on a board with tags, content, deadline, start date, assignee, parent task and custom fields supplied as code and value pairs
- Tasks retrieved across boards and filtered by name query, board, project, creator, assignee, status and Unix-timestamp ranges for deadline and creation date, with paging
- One task opened in full detail
- Projects listed and filtered, each carrying its boards — task boards come back with the tasks on them
- Jobs created in a workflow with the same shape of fields, and retrieved with filters that add a finish-date range
- One job opened in detail, and comments posted onto it
- Workflows listed with their stages and the field definitions used to create a job — name, code, type, whether it is required, and the available options
- User lookup returning username, id, name and a match score, which is how a person's name in a request becomes the id an assignment needs
Requirements
A Rework account and credentials from the platform settings — three access-token and password pairs, one per area: REWORK_PROJECT_ACCESS_TOKEN with REWORK_PROJECT_PASSWORD, REWORK_WORKFLOW_ACCESS_TOKEN with REWORK_WORKFLOW_PASSWORD, and REWORK_ACCOUNT_ACCESS_TOKEN with REWORK_ACCOUNT_PASSWORD. Run it with npx -y rework-mcp-server over stdio, from the npm package rework-mcp-server, currently 1.1.5. ENABLE_SSE=true switches it to an SSE server on PORT, which defaults to 3000, and there is a Docker Compose setup for the same. DISABLED_TOOLS takes a comma-separated list of tool names to leave unregistered, which is the lever when a client is straining under the number of tools. LOG_LEVEL controls log verbosity.
Setup effort
One command plus credentials — npx -y rework-mcp-server, then supply credentials
USE CASES
01
Filing a task from a conversation
Find the project and its board, then create the task with the deadline and assignee filled in.
02
Assigning work to a person you only know by name
Look the user up first to get their username and id, then pass those into the task you create.
03
Working out what is overdue
Filter tasks by deadline range and status across boards, then open the ones that need a closer look.
04
Starting a job in a workflow without guessing the fields
Read the workflow's stages and field definitions — which are required, what options they take — then create the job.
05
Recording progress on a job in flight
Open the job, then post the comment against it.
06
Reporting on what a workflow completed in a period
Filter jobs by finish-date range and status, paging through the results.
10 tools
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
create_job
Creates a single job in a Workflow. Use workflow_id. Required: name + workflow_id. Supports custom fields as array of {code, value}.
create_task
Creates a single task in a Project board. Use board_id. Required: name + board_id. Supports custom fields as array of {code, value}.
find_users
Find users from of Rework, returns an array of users contains username, id, name, score (0-100, higher is better). Every question about user (assignee, creator, etc.) should use this tool first to get username and user id and put it into the params if needed If you want an user, and not found any one yet, take the closest match.
get_detail_job
Get detail of a job by ID of job
get_detail_task
Get detail of a task by ID of task
get_jobs
Get jobs in the system. Can filtered by q (query to filter job name), deadline_from (Unix timestamp in seconds), deadline_to (Unix timestamp in seconds), created_from (Unix timestamp in seconds), created_to (Unix timestamp in seconds), finish_from (Unix timestamp in seconds), finish_to (Unix timestamp in seconds), page (page number), creator_username (username of creator), username (username of assignee), q (query to filter job name), workflow_id (ID of the workflow that the jobs belong to),
status (one of: 'active' (not done), 'done', 'failed', 'overdue'))
get_projects
Get projects. Can filtered by q. Each project contains list of boards with metatype tasks or docs. Tasks board will contains list of tasks
get_tasks
Get tasks from a all boards. Can filtered by q (query to filter task name), board_id, project_id, deadline_from (Unix timestamp in seconds), deadline_to (Unix timestamp in seconds), created_from (Unix timestamp in seconds), created_to (Unix timestamp in seconds), page (page number), creator (user id of creator), assign (user id of assignee), status (one of: 'active' (not done), 'done', 'review' (In review process), 'todo' (not started), 'doing' (started), 'donelate' (completed after deadline), 'overdue' (not done and past deadline), 'notreview' (not in review), 'today' (due today))
get_workflows
Get workflows. Can filtered by q. Each workflow contain a list of stages and list of fields that used to create job In each field, it contains:
- name: name of the field
- code: code of the field
- type: type of the field
- required: true if the field is required
- options: options of the field
- placeholder: description of the field
BEFORE YOU INSTALL
What to know first
Tools you get
10 tools when we connected; the project documents 11. list_all_users is documented but missing from a default install, where comment_job appears instead. DISABLED_TOOLS takes a comma-separated list of tools to remove when a client hits tool-count or context limits.
Requires
- npx on PATH
- REWORK_PROJECT_ACCESS_TOKEN in your environment
- REWORK_PROJECT_PASSWORD in your environment
- REWORK_WORKFLOW_ACCESS_TOKEN in your environment
- REWORK_WORKFLOW_PASSWORD in your environment
- REWORK_ACCOUNT_ACCESS_TOKEN in your environment
- REWORK_ACCOUNT_PASSWORD in your environment
Where it runs
The documented client runs the server on your machine.
INSTALLATION
Connect your client
Server setup
TERMINAL
npx -y rework-mcp-serverMCP Clients
Any MCP client — npx
ANY MCP CLIENT — NPX · JSON
{
"mcpServers": {
"Rework": {
"command": "npx",
"args": [
"-y",
"rework-mcp-server"
],
"env": {
"REWORK_PROJECT_ACCESS_TOKEN": "",
"REWORK_PROJECT_PASSWORD": "",
"REWORK_WORKFLOW_ACCESS_TOKEN": "",
"REWORK_WORKFLOW_PASSWORD": "",
"REWORK_ACCOUNT_ACCESS_TOKEN": "",
"REWORK_ACCOUNT_PASSWORD": ""
}
}
}
}
