harvest-forecast-mcp
An MCP server that integrates Harvest time tracking and Forecast resource scheduling APIs, enabling AI agents to track time, manage projects, and handle assignments and scheduling.
README
harvest-forecast-mcp
An MCP server for the Harvest and Forecast APIs — let AI agents like Claude track time, list projects, and manage assignments and scheduling.
What it does
harvest-forecast-mcp exposes Harvest (time tracking) and Forecast (resource scheduling) operations as Model Context Protocol tools. Once configured, an AI agent can:
- Track time — list, create, and delete time entries in Harvest
- Browse projects — list projects, clients, tasks, and users
- Manage scheduling — list and create Forecast assignments, milestones, and holidays
The server speaks stdio MCP, so it drops cleanly into Claude Desktop, Claude Code, or any MCP-compatible client.
Installation
# Run instantly without installing (recommended)
uvx harvest-forecast-mcp
# Or install with pip
pip install harvest-forecast-mcp
Configuration
The server reads credentials from environment variables. Harvest and Forecast use separate access tokens and account IDs, so you can configure one or both depending on which tools you need.
| Variable | Required | Description |
|---|---|---|
HARVEST_ACCESS_TOKEN |
Harvest tools | Personal access token for the Harvest API |
HARVEST_ACCOUNT_ID |
Harvest tools | Harvest account ID (numeric) |
FORECAST_ACCESS_TOKEN |
Forecast tools | Personal access token for the Forecast API |
FORECAST_ACCOUNT_ID |
Forecast tools | Forecast account ID (numeric) |
HARVEST_USER_AGENT |
Yes | Identifying string for API requests (e.g. your-app (you@example.com)) |
Tip: Harvest requires a descriptive
User-Agentheader. SetHARVEST_USER_AGENTto your app name and contact email.
Claude Desktop configuration
Add the server to your claude_desktop_config.json:
{
"mcpServers": {
"harvest-forecast": {
"command": "uvx",
"args": ["harvest-forecast-mcp"],
"env": {
"HARVEST_ACCESS_TOKEN": "your-harvest-token",
"HARVEST_ACCOUNT_ID": "123456",
"FORECAST_ACCESS_TOKEN": "your-forecast-token",
"FORECAST_ACCOUNT_ID": "123456",
"HARVEST_USER_AGENT": "my-claude-agent (you@example.com)"
}
}
}
}
Tools
The server exposes 28 tools across the Harvest and Forecast APIs, including cross-API combinations:
Harvest — time tracking & project management
| Tool | Description |
|---|---|
list_harvest_projects |
List all projects with client info, optionally filtered by active status |
list_harvest_users |
List all users, optionally filtered by active status |
list_harvest_clients |
List all clients, optionally filtered by active status |
list_harvest_tasks |
List all tasks available for time tracking |
list_time_entries |
List time entries filtered by user, project, and date range (defaults to last 30 days) |
create_time_entry |
Log hours against a project + task with optional notes |
list_project_assignments |
List users assigned to a Harvest project with rates and budgets |
harvest_whoami |
Get the current authenticated Harvest user's identity and permissions |
Forecast — scheduling & staffing
| Tool | Description |
|---|---|
list_forecast_projects |
List all Forecast projects (for scheduling) |
list_forecast_people |
List all schedulable people with roles and teams |
list_forecast_placeholders |
List all placeholders (role-based slots) |
list_forecast_roles |
List all roles with their assigned people and placeholders |
list_forecast_milestones |
List all milestones across all projects |
get_forecast_project |
Get details for a single Forecast project by ID |
get_forecast_person |
Get details for a single Forecast person by ID |
list_assignments |
List scheduled assignments filtered by date range, project, or person |
create_assignment |
Schedule a person on a project with allocation and dates |
update_assignment |
Update an existing assignment |
delete_assignment |
Delete an assignment |
get_remaining_budgeted_hours |
Get remaining budgeted hours for all projects |
get_future_scheduled_hours |
Get future scheduled hours from a date |
get_future_scheduled_hours_for_project |
Same, filtered to a single project |
get_assigned_people |
Get a mapping of project IDs to assigned person IDs |
get_project_heatmap |
Scheduling load heatmap for a project |
get_person_heatmap |
Scheduling load heatmap for a person |
get_placeholder_heatmap |
Scheduling load heatmap for a placeholder |
forecast_whoami |
Get current Forecast user identity and account IDs |
Cross-API — manager insights
| Tool | Description |
|---|---|
person_schedule |
A person's complete schedule: Forecast assignments + Harvest time entries for the period |
project_overview |
Comprehensive project view: Harvest details + Forecast project + budget + recent time entries |
team_utilization |
Scheduled hours (Forecast) vs logged hours (Harvest) per person, with utilization % |
who_is_available |
Find people with capacity (not fully scheduled) in a date range |
time_summary |
Summarize time entries grouped by project, user, or client |
budget_status |
Budget burn-down: remaining budgeted hours vs logged hours last 30 days per project |
all_projects_schedule |
Per-project scheduling overview: people, assignments, milestones for a date range |
Development
# Install dependencies
uv sync --group dev
# Run tests
uv run pytest
# Type check
uv run pyright src/harvest_forecast_mcp
# Lint
uv run ruff check .
# Format
uv run ruff format .
See CONTRIBUTING.md for full development guidelines.
License
MIT
Disclaimer
This project is unofficial and not affiliated with, endorsed by, or sponsored by Harvest Inc. or its subsidiaries. "Harvest" and "Forecast" are trademarks of their respective owners. This software is provided "as-is", without any warranty express or implied.
MIT — Copyright (c) 2025 AZX, PBC.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。