Mello Python SDK

Typed Python client for the Mello Public REST API.

The SDK wraps Mello workspaces, boards, columns, tickets, comments, history, and search endpoints with dataclass models and typed exceptions.

Installation

uv add mello-sdk

Install from source during development:

git clone <repository-url>
cd mello-python-sdk
uv sync --extra dev

If you are not using uv, install with pip install mello-sdk.

Quick Start

from mello import MelloClient

client = MelloClient(token="YOUR_PERSONAL_API_TOKEN")

user = client.get_current_user()
print(f"Logged in as {user.name} ({user.email})")

for workspace in client.list_workspaces():
    print(f"Workspace: {workspace.name}")

    for board in client.list_workspace_boards(workspace.id):
        print(f"  Board: {board.name} ({board.code})")

CLI

mello-cli is a JSON-first command-line interface for scripts and AI agents. It uses MELLO_API_KEY by default and emits exactly one JSON object on stdout.

export MELLO_API_KEY="mello_pat_..."

mello-cli me get
mello-cli workspace list
mello-cli ticket search --workspace-id "workspace-uuid" --query "login crash"
mello-cli ticket update --ticket-id "ticket-uuid" --set title="Fix login on iOS"
mello-cli ticket update --ticket-id "ticket-uuid" --clear pic_user_id

Omitted update fields stay unchanged. --clear field (or --set field=null) clears nullable fields such as pic_user_id, supervisor_id, and dates.

Destructive and high-impact operations require explicit confirmation. After confirming the target with the user, pass --yes:

mello-cli --yes ticket delete --ticket-id "ticket-uuid"
mello-cli --yes github replace-board-repos \
  --workspace-id "workspace-uuid" --board-id "board-uuid" \
  --repositories '[{"installation_id": 1, "github_repo_id": 2}]'

Use --token, --base-url, and --timeout to override MELLO_API_KEY, MELLO_BASE_URL, and MELLO_TIMEOUT. Run mello-cli --help for the complete resource command surface.

Install the CLI globally

Install mello-cli (and mello-mcp-server) globally as an isolated tool with uv:

# From PyPI
uv tool install mello-sdk

# Or directly from this repository (local development)
uv tool install --from . mello-sdk

The executables land in ~/.local/bin (make sure it is on your PATH).

Important: when installed from a local source directory, the global tool does not track your changes. After updating the CLI code in this repo, reinstall to refresh the global executables:

uv tool install --from . mello-sdk --force

Claude Code Skill

The repository ships a Claude Code skill at skills/mello/ that teaches agents to operate Mello through mello-cli (JSON output, safe confirmation rules, update semantics).

To use it, install the skill into your Claude Code configuration:

# Personal (available in every project)
cp -r skills/mello ~/.claude/skills/mello

# Or project-local (only inside a specific project)
cp -r skills/mello /path/to/your-project/.claude/skills/mello

Then make sure mello-cli is installed globally (see above) and MELLO_API_KEY is set in your shell environment. Claude Code will trigger the mello skill automatically for Mello-related requests.

MCP Server

The package can also run as a Model Context Protocol server for AI assistants that support MCP tools.

Install the MCP extra:

uv add "mello-sdk[mcp]"

For local development from this repository:

uv sync --extra mcp --extra dev

Configure the server with environment variables:

export MELLO_API_KEY="mello_pat_..."
export MELLO_BASE_URL="https://mello.mezon.vn/api/v1"  # optional
export MELLO_TIMEOUT="30"                              # optional seconds

Run it with the console script:

uv run mello-mcp-server

Or run the module directly:

uv run python -m mello.mcp_server

The MCP server exposes the SDK's full read/write surface: workspaces, boards, columns, tickets, comments, history, and search. Update tools accept an updates object so omitted fields are left unchanged while explicit null values are sent to Mello for nullable fields.

Transport

main() selects the transport from the MCP_TRANSPORT environment variable. The default is stdio for local assistant integrations. Set it to streamable-http (or sse) to expose the server over HTTP. In HTTP mode the bind address is controlled by MCP_HOST (default 0.0.0.0) and MCP_PORT (default 8000).

MCP_TRANSPORT=streamable-http MCP_PORT=8000 uv run mello-mcp-server

Docker

The repository ships a Dockerfile and docker-compose.yml that run the server with the streamable-http transport on port 8000.

Build and run with Docker:

docker build -t mello-mcp-server .
docker run --rm -p 8000:8000 -e MELLO_API_KEY="mello_pat_..." mello-mcp-server

Or use Docker Compose (reads MELLO_API_KEY from your environment or .env):

export MELLO_API_KEY="mello_pat_..."
docker compose up --build

The HTTP endpoint is served at http://localhost:8000/mcp. Point an MCP client that supports the streamable-http transport at that URL.

Usage

Boards

board = client.create_board(
    workspace_id="workspace-uuid",
    name="Q3 Planning",
    code="Q3PL",
)

board_detail = client.get_board(board.id)

client.update_board(
    board.id,
    name="Q3 Project Planning",
    background_color="#3b5998",
)

client.delete_board(board.id)

Columns

column = client.create_column(
    board_id="board-uuid",
    name="In Review",
    position=2,
)

client.update_column(column.id, name="Code Review", color="#ffcc00")

client.reorder_columns(
    board_id="board-uuid",
    column_ids=["column-uuid-1", "column-uuid-2", "column-uuid-3"],
)

Tickets

ticket = client.create_ticket(
    column_id="column-uuid",
    title="Fix login crash",
    description="Steps to reproduce...",
)

ticket_detail = client.get_ticket(ticket.id)
print(len(ticket_detail.comments))

client.update_ticket(
    ticket.id,
    title="Fix login crash on iOS",
    pic_user_id="user-uuid",
)

# Nullable fields can be cleared explicitly with None.
client.update_ticket(
    ticket.id,
    pic_user_id=None,
    supervisor_id=None,
    start_date=None,
    end_date=None,
)

client.move_ticket(ticket.id, column_id="other-column-uuid", position=0)

Comments, History, And Search

comment = client.create_comment(
    ticket_id="ticket-uuid",
    body="Investigating this issue now.",
)

comments = client.list_comments(ticket_id="ticket-uuid")
history = client.list_history(ticket_id="ticket-uuid")
results = client.search_tickets(workspace_id="workspace-uuid", q="login crash")

Error Handling

The SDK raises typed exceptions derived from MelloAPIException for API errors:

from mello import (
    MelloAPIException,
    MelloClient,
    ForbiddenException,
    NotFoundException,
    RateLimitedException,
    UnauthorizedException,
    ValidationErrorException,
)

client = MelloClient(token="YOUR_PERSONAL_API_TOKEN")

try:
    client.get_current_user()
except UnauthorizedException:
    print("Invalid or expired API token.")
except ForbiddenException:
    print("The token cannot access this resource.")
except NotFoundException:
    print("Resource not found.")
except ValidationErrorException as exc:
    print(f"Validation failed: {exc.fields}")
except RateLimitedException:
    print("Rate limit exceeded. Try again later.")
except MelloAPIException as exc:
    print(f"Mello API error {exc.status_code}: {exc.error_code}")

Development

Install development dependencies:

python3 -m venv .venv
source .venv/bin/activate
pip install -e ".[dev]"

Run unit tests:

pytest

Run live integration tests. These tests create and delete data in your Mello workspace, so use a dedicated test token when possible:

export MELLO_API_KEY="mello_pat_..."
pytest -m integration

Run quality checks:

black --check mello tests
flake8 mello tests
mypy mello tests

Build And Publish

The package uses pyproject.toml with setuptools. Build artifacts locally:

python -m build
python -m twine check dist/*

Recommended release flow:

rm -rf dist/ build/ *.egg-info
python -m build
python -m twine check dist/*
python -m twine upload --repository testpypi dist/*

After validating installation from TestPyPI, publish to PyPI:

python -m twine upload dist/*

Use PyPI API tokens instead of passwords, and avoid committing .env, .pypirc, dist/, or build artifacts.

License

MIT. See LICENSE.