mcptask.online

online.mcptaskPublished by mcptask.onlineOfficial publisherv1.0.2Updated Nov 12, 2026

Autonomous AI developers on your own infrastructure: assign a task, get a pull request.

VerifiedStreamable HTTPWeb executableDeveloper ToolsAI & ML

Overview

AI-generated overview

Remote MCP server for mcptask.online that lets an assistant assign coding tasks to autonomous AI developers and get pull requests back.

What you can do
  • Assign development tasks to autonomous AI developers running on your own infrastructure
  • Retrieve the resulting pull requests from the task service
  • Connect from Claude Code, Claude.ai, ChatGPT or any other MCP client over Streamable HTTP
  • Work alongside the separate mcptask runner binary, which drives Claude Code through the same tasks
When to use it
Useful when you want an assistant to hand off coding work to autonomous agents on your own infrastructure and review the pull requests they produce, rather than editing code directly in the conversation.
Requirements
A remote endpoint at mcptask.online/mcp over Streamable HTTP. Authentication is required: OAuth, or a bearer token obtained from the service and supplied in the Authorization header (for example via an MCPTASK_TOKEN environment variable in the client config). A mcptask.online account is implied.
Before you install
Access is gated by OAuth or a bearer token, so the token is a credential to protect. The server assigns work to autonomous agents that produce pull requests, so changes to your repositories are expected; review them before merging. The README notes the runner binary runs Claude sessions with your credentials, and the project is proprietary.

Installation

In SourceWeft

  1. Open mcptask.online in the dashboard and add it to a workspace.
  2. Connect the required credentials. They are stored privately and never shown on public pages.
  3. Enable the server for the chats that should use its tools.

Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.

Use with an agent

Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.

Read https://sourceweft.com/mcp/online.mcptask%2Fmcptask/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.

Read the install guide the agent follows

Other MCP clients

Add this to your client's mcpServers config. The client completes OAuth sign-in on first connect.

{
  "mcpServers": {
    "mcptask-online": {
      "type": "http",
      "url": "https://mcptask.online/mcp"
    }
  }
}

README

mcptask-releases

Released binaries of the mcptask runner — the tool that drives Claude Code through the tasks on mcptask.online.

This repository holds no source. It exists so the binaries can be downloaded without a token: the runner is developed in a private repository, and a private repository's release assets cannot be fetched anonymously, which would rule out a one-line install. The artefacts are the public part.

Install

bash
curl -fsSL https://github.com/jchsoft/mcptask-releases/releases/latest/download/install.sh | sh

macOS and Linux, amd64 and arm64. The script detects the platform, verifies the download against the release's checksums.txt, and installs mcptask_runner into /usr/local/bin when that is already writable and ~/.local/bin otherwise. It never calls sudo, and it installs the binary and stops — it does not run init and does not enable a scheduled job.

Read it before you run it: install.sh is the same file the command above fetches.

VariableEffect
MCPTASK_VERSIONinstall a specific tag instead of the latest release
MCPTASK_INSTALL_DIRinstall somewhere else

On Windows, download the windows archive from Releases and put mcptask_runner.exe on PATH.

The feature catalog

catalog.yml is the runner's feature catalog in English, Czech and Slovak: one entry per feature, subcommand, flag, config key, harness, git host and exit code. It is generated in the private source repository and validated there by a test that fails when the catalog and the code disagree, and every release carries it as an asset. mcptask.online/runner renders it. The latest one is always at https://github.com/jchsoft/mcptask-releases/releases/latest/download/catalog.yml and a specific version's at https://github.com/jchsoft/mcptask-releases/releases/download/<tag>/catalog.yml.

Verifying a manual download

Every release publishes checksums.txt alongside the archives.

bash
shasum -a 256 --ignore-missing --check checksums.txt   # sha256sum -c on Linux

Worth doing: the binary goes on to run your Claude sessions with your credentials.

Getting started

bash
mcptask_runner init      # in the project you want the runner to work onmcptask_runner version

MCP server

[smithery badge] [Glama connector]

mcptask.online is itself a remote MCP server — the runner is one client of it, Claude Code, Claude.ai, ChatGPT and any other MCP client can connect directly.

Endpointhttps://mcptask.online/mcp (Streamable HTTP)
AuthenticationAuthorization: Bearer <token> or OAuth; without a token /mcp answers 401 with WWW-Authenticate
Server card/.well-known/mcp/server-card.json — every tool and resource
Registry entryserver.json (online.mcptask/mcptask)

Put this in the project's .mcp.json and export your token as MCPTASK_TOKEN:

json
{  "mcpServers": {    "mcptask-online": {      "type": "http",      "url": "https://mcptask.online/mcp",      "headers": {        "Authorization": "Bearer ${MCPTASK_TOKEN}"      }    }  }}

Where to get the token and how to wire up each client: mcptask.online/installation.

Issues and questions: mcptask.online. Proprietary — © JCHSoft. All rights reserved.

Source: README.md at commit 2e827dd

Tools

0
This server asks you to sign in, so SourceWeft could not read its tools without an account and has not indexed them. What a connected account sees can depend on that account and its permissions.

Version history

3
  1. v1.0.2LatestNov 12, 2026
  2. v1.0.1Oct 23, 2026
  3. v1.0.0Sep 28, 2026

Score

75% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 daysRequired25 pts
  • Auth declaredRequired10 pts
  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts

    Declare a license in the repository or the registry entry.