Freedcamp MCP Server
Enables to manage Freedcamp projects, tasks, users, and comments through natural language with HMAC-SHA1 authentication and name resolution.
README
Freedcamp MCP Server
A Model Context Protocol server that wraps the Freedcamp REST API. It lets any MCP-compatible LLM client (Claude Code, Claude Desktop, etc.) manage Freedcamp projects, tasks, users, and comments through natural language.
Features
- 17 tools covering projects, tasks, users, comments, and health checks
- HMAC-SHA1 authentication — the API secret never leaves the server; only a signed hash is sent per request
- Name resolution — pass user names, emails, or project names instead of raw numeric IDs; the server resolves them automatically with TTL-based caching
- Field limiting — request only the fields you need with dot-notation (
id,title,comments.created_ts) to reduce response size - Status label mapping — accept human-readable strings like
"in progress"instead of numeric codes - Response filtering — internal API fields are automatically stripped from responses
- Graceful shutdown — drains in-flight requests before exiting
- Retry + backoff — retries on 429 and 5xx with exponential backoff
- No build step — runs TypeScript directly via tsx
Prerequisites
- Node.js >= 18
- A Freedcamp account with API credentials (Settings → API)
Installation
git clone https://github.com/mahrukh-n8n/freedcampMCP.git
cd freedcampMCP
npm install
Configuration
Option A: .env file
cp .env.example .env
# Edit .env with your Freedcamp API key and secret
Option B: Claude Code MCP settings
No .env file needed — pass credentials as environment variables:
claude mcp add freedcamp npx tsx /path/to/freedcampMCP/scripts/mcp-server.ts \
-e FREEDCAMP_API_KEY=your_key \
-e FREEDCAMP_API_SECRET=your_secret
Environment Variables
| Variable | Required | Default | Description |
|---|---|---|---|
FREEDCAMP_API_KEY |
Yes | — | Freedcamp API key |
FREEDCAMP_API_SECRET |
Yes | — | Freedcamp API secret |
FREEDCAMP_API_URL |
No | https://freedcamp.com |
Base URL (for self-hosted) |
LOG_LEVEL |
No | info |
Log level: debug, info, warn, error |
REQUEST_TIMEOUT_MS |
No | 30000 |
HTTP request timeout (ms) |
CACHE_TTL_MS |
No | 60000 |
Name resolution cache TTL (ms) |
MAX_CONCURRENT_REQUESTS |
No | 6 |
Max concurrent API requests |
Running
With Claude Code (recommended)
After adding the MCP server with claude mcp add, just start a conversation. Claude will call tools automatically when needed.
With MCP Inspector
npx @modelcontextprotocol/inspector npx tsx scripts/mcp-server.ts
Opens a browser UI where you can call each tool and inspect responses.
Direct (stdio)
npx tsx scripts/mcp-server.ts
The server listens on stdin/stdout using the MCP stdio transport. The host process (Claude Code, Claude Desktop) manages its lifecycle.
Tools
Health
| Tool | Description |
|---|---|
health.check |
Verify API credentials and connection status |
Projects
| Tool | Write | Description |
|---|---|---|
project.list |
List projects (paginated, sortable, field-limitable) | |
project.get |
Get project by ID or name | |
project.create |
Yes | Create a project (name, description, color, group, members) |
project.update |
Yes | Update project fields (partial update) |
Tasks
| Tool | Write | Description |
|---|---|---|
task.list |
List tasks with filters (assignee, status, date range, search, tags) | |
task.get |
Get task by ID with comments and tag detail; injects task_url |
|
task.create |
Yes | Create task (status labels accepted, file attachments) |
task.update |
Yes | Update task fields (partial update, file attachments) |
task.delete |
Yes | Delete a task |
task.assign |
Yes | Assign users to a task |
Users
| Tool | Write | Description |
|---|---|---|
user.list |
List users (optionally filter by project) | |
user.get |
Get user by ID, email, or name | |
user.current |
Get authenticated user's profile | |
user.create |
Yes | Create user (email, password, first_name, OAuth) |
user.update_current |
Yes | Update authenticated user's profile |
Comments
| Tool | Write | Description |
|---|---|---|
comment.add |
Yes | Add a comment (requires item_id + app_id) |
comment.update |
Yes | Update comment text |
comment.delete |
Yes | Delete a comment |
Name Resolution
Most ID parameters accept names, emails, or numeric IDs. Examples:
project_id: "Marketing"— resolves to the project's numeric IDassigned_to_id: "alice@example.com"— resolves to the user's numeric IDassigned_to_id: ["Alice", 42]— mixed lists accepted
Resolution results are cached with a configurable TTL (CACHE_TTL_MS).
Status Mapping
Task status accepts both numeric codes and string labels:
| Code | Label |
|---|---|
| 0 | not started |
| 1 | in progress |
| 2 | completed |
Example: status: "in progress" is equivalent to status: 1.
Field Limiting
All list and get tools accept a fields parameter with dot-notation paths:
fields="id,title,priority,comments.created_ts"
This reduces response size and focuses the LLM on relevant data. Nested arrays are preserved — comments.created_ts on [{created_ts: 1}] yields [{created_ts: 1}], not a flat list.
App ID Constants (for comments)
| App | ID |
|---|---|
| tasks | 2 |
| milestones | 3 |
| discussions | 5 |
| files | 6 |
| time | 8 |
| issue_tracker | 9 |
Authentication
The server uses HMAC-SHA1 authentication. On every request:
- A Unix timestamp is generated
- A hash is computed:
HMAC-SHA1(secret, apiKey + timestamp) - Auth params are sent as query string:
?api_key=...×tamp=...&hash=...
The secret never goes over the wire. At boot, the server validates credentials with GET /api_key/check.
Error Codes
| Code | Meaning |
|---|---|
PERMISSION_DENIED |
Invalid API key/secret or insufficient access |
NOT_FOUND |
Requested resource or name resolution target does not exist |
VALIDATION_ERROR |
Invalid input parameters |
CONFLICT |
Resource already exists |
INTERNAL_ERROR |
Server error, rate limit, or network failure |
Development
# Type check
npx tsc --noEmit
# Run tests
npx vitest run
# Watch mode
npx vitest
# Run server in dev mode
npm run dev
Testing
The test suite uses Vitest with mocked API responses:
npx vitest run # Single run
npx vitest # Watch mode
npx vitest --coverage # With coverage
Project Structure
scripts/mcp-server.ts Entry point
src/lib/freedcamp/
api-client.ts HTTP client with HMAC auth, retry, filtering
register-tools.ts Wire all tools to the MCP registry
auth/hmac.ts HMAC-SHA1 computation
auth/hmac-validator.ts Boot-time credential validation
tools/
health.ts health.check
projects.ts project.list/get/create/update
tasks.ts task.list/get/create/update/delete/assign
users.ts user.list/get/current/create/update_current
comments.ts comment.add/update/delete
utils/
name-resolver.ts Name/email → ID resolution with caching
response-filter.ts Strip internal fields from API responses
field-limiter.ts Dot-notation field extraction
date-utils.ts Date validation and formatting
resolution-cache.ts TTL-based LRU cache
logger.ts Structured logging with verbose mode
validation.ts Input validation helpers
src/modules/mcp/
registry/tool-registry.ts MCP tool registry
services/create-mcp-server.ts MCP server factory
services/stdio-transport.ts Stdio transport
types.ts MCP result types
utils/serialize.ts Result envelope helpers (dataResult, commitResult, etc.)
License
MIT
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。