productplan-mcp-server
ProductPlan MCP 服务器
通过 AI 与您的路线图对话。 通过与 Claude、Cursor 或其他 AI 助手进行自然对话,您可以提问、创建想法、检查 OKR 进度并管理发布。
您可以用它做什么?
无需点击 ProductPlan 的界面,只需询问:
“我们第一季度的路线图上有什么?”
“向我展示所有进度落后的目标”
“为移动应用改进创建一个新想法”
“本月有哪些发布计划?”
“列出所有标记为 'customer-request' 的想法”
AI 会获取您真实的 ProductPlan 数据并在几秒钟内做出响应。
Related MCP server: plane-mcp-server
这是为谁准备的?
产品经理:希望更快地访问路线图数据
团队负责人:需要快速获取状态更新而无需切换上下文
任何使用 AI 助手(Claude、Cursor 等)并希望将 ProductPlan 集成到工作流中的用户
无需编码。您只需复制一个文件并粘贴一些设置即可。
快速入门(5 分钟)
第 1 步:获取您的 ProductPlan API 令牌
登录 ProductPlan
转到 Settings → API(或直接访问 此链接)
复制您的 API 令牌
第 2 步:下载应用程序
前往 Releases 页面 并下载适合您计算机的文件:
您的计算机 | 下载此文件 |
Mac (M1, M2, M3, M4) |
|
Mac (Intel) |
|
Windows |
|
Linux |
|
在 Mac/Linux 上,打开终端并运行这两条命令(将文件名替换为您下载的文件名):
chmod +x ~/Downloads/productplan-darwin-arm64
sudo mv ~/Downloads/productplan-darwin-arm64 /usr/local/bin/productplan系统会要求您输入密码。这是正常的。
在 Windows 上:
为二进制文件创建一个文件夹(如果不存在):
mkdir C:\Tools将下载的
.exe移动到该文件夹并重命名:move %USERPROFILE%\Downloads\productplan-windows-amd64.exe C:\Tools\productplan.exe在您的 AI 助手配置中使用完整路径
C:\Tools\productplan.exe(在第 3 步中显示)
注意:您可以跳过添加到 PATH 的步骤。只需在配置中使用完整文件路径即可。
第 3 步:连接到您的 AI 助手
选择您使用的工具:
找到您的配置文件:
Mac:
~/Library/Application Support/Claude/claude_desktop_config.jsonWindows:
%APPDATA%\Claude\claude_desktop_config.json
在任何文本编辑器中打开它并添加以下内容(将
your-token替换为您实际的 API 令牌):
Mac/Linux:
{
"mcpServers": {
"productplan": {
"command": "/usr/local/bin/productplan",
"env": {
"PRODUCTPLAN_API_TOKEN": "your-token"
}
}
}
}Windows:
{
"mcpServers": {
"productplan": {
"command": "C:\\Tools\\productplan.exe",
"env": {
"PRODUCTPLAN_API_TOKEN": "your-token"
}
}
}
}重启 Claude Desktop
添加到您的配置文件:
Mac/Linux:
~/.claude.jsonWindows:
%USERPROFILE%\.claude.json
Mac/Linux:
{
"mcpServers": {
"productplan": {
"command": "/usr/local/bin/productplan",
"env": {
"PRODUCTPLAN_API_TOKEN": "your-token"
}
}
}
}Windows:
{
"mcpServers": {
"productplan": {
"command": "C:\\Tools\\productplan.exe",
"env": {
"PRODUCTPLAN_API_TOKEN": "your-token"
}
}
}
}打开 Cursor
转到 Settings → MCP Servers
添加此配置:
Mac/Linux:
{
"productplan": {
"command": "/usr/local/bin/productplan",
"env": {
"PRODUCTPLAN_API_TOKEN": "your-token"
}
}
}Windows:
{
"productplan": {
"command": "C:\\Tools\\productplan.exe",
"env": {
"PRODUCTPLAN_API_TOKEN": "your-token"
}
}
}Windows 用户:在路径中使用双反斜杠 (
\\)。这是必需的,因为反斜杠在 JSON 中是转义字符。
安装 Cline 扩展
打开 VS Code 设置 (JSON) 并添加:
Mac/Linux:
{
"cline.mcpServers": {
"productplan": {
"command": "/usr/local/bin/productplan",
"env": {
"PRODUCTPLAN_API_TOKEN": "your-token"
}
}
}
}Windows:
{
"cline.mcpServers": {
"productplan": {
"command": "C:\\Tools\\productplan.exe",
"env": {
"PRODUCTPLAN_API_TOKEN": "your-token"
}
}
}
}安装 Continue 扩展
添加到您的配置文件:
Mac/Linux:
~/.continue/config.jsonWindows:
%USERPROFILE%\.continue\config.json
Mac/Linux:
{
"mcpServers": [
{
"name": "productplan",
"command": "/usr/local/bin/productplan",
"env": {
"PRODUCTPLAN_API_TOKEN": "your-token"
}
}
]
}Windows:
{
"mcpServers": [
{
"name": "productplan",
"command": "C:\\Tools\\productplan.exe",
"env": {
"PRODUCTPLAN_API_TOKEN": "your-token"
}
}
]
}在您的 n8n 实例上设置环境变量:
N8N_COMMUNITY_PACKAGES_ALLOW_TOOL_USAGE=true将 MCP Client 节点添加到您的工作流
配置:
Command:
Mac/Linux:
/usr/local/bin/productplanWindows:
C:\Tools\productplan.exe
Environment Variables:
PRODUCTPLAN_API_TOKEN=your-token
连接到 AI Agent 节点
示例工作流:Slack Trigger → AI Agent (with MCP Client) → Slack Response
第 4 步:开始提问
打开您的 AI 助手并尝试:
“列出我的 ProductPlan 路线图”
“路线图 [名称] 上有哪些条目?”
“向我展示我们的 OKR”
“哪些想法处于发现阶段?”
实际用例
早会准备
“总结过去一周我们产品路线图上的变化”
利益相关者更新
“列出所有第一季度的目标及其进度”
想法分类
“向我展示所有标记为 'enterprise' 且未设置优先级的想法”
发布协调
“一月份的发布还有哪些任务未完成?”
快速查找
“'Mobile App v2' 条目计划何时开始?”
您可以访问哪些 ProductPlan 数据?
功能 | 查看 | 创建 | 编辑 | 删除 |
路线图 | 是 | - | - | - |
路线图评论 | 是 | - | - | - |
条目 (路线图项) | 是 | 是 | 是 | 是 |
条目评论 | 是 | - | - | - |
条目连接 | 是 | 是 | - | 是 |
条目链接 | 是 | 是 | - | 是 |
泳道 (类别) | 是 | 是 | 是 | 是 |
图例 (条目颜色) | 是 | - | - | - |
里程碑 | 是 | 是 | 是 | 是 |
想法 (发现) | 是 | 是 | 是 | - |
想法客户 | 是 | - | - | - |
想法标签 | 是 | - | - | - |
机会 | 是 | 是 | 是 | - |
想法表单 | 是 | - | - | - |
目标 (OKR) | 是 | 是 | 是 | 是 |
关键结果 | 是 | 是 | 是 | 是 |
发布 | 是 | 是 | 是 | 是 |
发布部分 | 是 | 是 | 是 | 是 |
发布任务 | 是 | 是 | 是 | 是 |
用户 | 是 | - | - | - |
团队 | 是 | - | - | - |
工作原理
┌─────────────────┐ spawns ┌─────────────────┐ API calls ┌─────────────────┐
│ AI Assistant │ ───────────────── │ MCP Server │ ─────────────────▶ │ ProductPlan │
│ (Claude, Cursor)│ ◀───────────────▶ │ (this binary) │ ◀───────────────── │ API │
└─────────────────┘ stdin/stdout └─────────────────┘ JSON data └─────────────────┘
your computer your computer cloud为什么这需要在您的计算机上运行?
MCP (Model Context Protocol) 通过子进程模型工作。您的 AI 助手不会连接到远程服务器;它将二进制文件作为本地进程启动,并通过 stdin/stdout 进行通信。这种架构意味着:
二进制文件必须存在于本地,因为您的 AI 助手将其作为子进程运行
您的 API 令牌保留在您的机器上,绝不会通过第三方服务器
实时、同步通信,AI 和 MCP 服务器之间没有网络延迟
离线工作(针对缓存数据,尽管 ProductPlan API 调用仍需要互联网)
当您询问“我们第一季度的路线图上有什么?”时,会发生以下情况:
您的 AI 助手识别出它需要 ProductPlan 数据
它向 MCP 服务器进程发送结构化请求
二进制文件将其转换为 ProductPlan API 调用
ProductPlan 返回 JSON 数据
二进制文件格式化并将结果返回给您的 AI
您的 AI 以自然语言呈现答案
代理技能
预构建的工作流指南,教导 AI 助手如何有效地使用 ProductPlan 工具。每项技能都针对具有量身定制工作流的特定角色。
技能 | 受众 | 重点 |
通用 | 核心模式和工具参考 | |
产品经理 | 全套工具:路线图、OKR、想法、发布 | |
高管 | 组合健康状况、跨路线图视图 | |
销售与客户成功 | 面向客户的路线图时间表 |
共享原则
所有技能都遵循这些输出约定:
无原始 JSON - 将响应格式化为可读的文本和表格
人类可读的日期 - 使用 “2025 年 3 月” 或 “2025 年第一季度”,而不是 “2025-03-15”
总结大型列表 - 不要用 50 个项目压倒用户;提供展开选项
角色特定的变体:
PM 包含
bar_id以便进行后续操作领导层 以执行摘要开头,隐藏实现细节
面向客户 完全省略内部 ID、泳道名称和 OKR
要使用技能,请将 SKILL.md 文件复制到您的 Claude Code 技能目录:
# Copy a skill (example: PM skill)
cp skills/productplan-pm/SKILL.md ~/.claude/skills/productplan-pm.md或者在您的提示词中直接引用技能:
“使用 productplan-pm 工作流向我展示我们的第一季度路线图”
故障排除
“Command not found” 或 “spawn ENOENT”
您的 AI 助手找不到二进制文件。这意味着:
Mac/Linux: 文件不在
/usr/local/bin/productplan,或者您忘记运行chmod +xWindows: 配置中的路径与您保存
.exe的位置不匹配
解决方法:验证二进制文件是否存在于配置中的路径。运行 ls -la /usr/local/bin/productplan (Mac/Linux) 或检查 C:\Tools\productplan.exe 是否存在 (Windows)。
Windows 路径问题
Windows 上的常见错误:
错误 | 正确 |
|
|
|
|
|
|
缺少 | 在路径中包含 |
Windows 使用反斜杠 (\) 作为路径,但 JSON 将反斜杠视为转义字符。您必须在配置文件中将它们加倍 (\\)。
“Invalid API token”
请在 ProductPlan Settings → API 仔细检查您的令牌。令牌可能会过期或被重新生成。确保您复制了完整的令牌,没有多余的空格。
“No roadmaps found”
您的 API 令牌只能访问您在 ProductPlan 中有权查看的数据。检查您的帐户是否拥有您正在查找的路线图的访问权限。
AI 助手看不到 ProductPlan 工具
MCP 服务器在您的 AI 助手启动时加载,而不是在配置更改时加载。编辑配置文件后,请完全退出并重启应用程序。在 Mac 上,使用 Cmd+Q(不仅仅是关闭窗口)。
“Permission denied” (Mac/Linux)
二进制文件需要执行权限。运行:
chmod +x /usr/local/bin/productplan命令行 (可选)
您也可以直接在终端中使用此工具,无需 AI 助手:
# First, set your token
export PRODUCTPLAN_API_TOKEN="your-token"
# Then run commands
productplan status # Check connection
productplan roadmaps # List all roadmaps
productplan bars 12345 # List bars in roadmap #12345
productplan objectives # List all OKRs
productplan ideas # List all ideas
productplan opportunities # List all opportunities
productplan launches # List all launches背景信息
什么是 MCP?
Model Context Protocol (MCP) 是一种开放标准,允许 AI 助手连接到外部工具。Anthropic 创建了它;其他 AI 提供商正在采用它。此服务器实现了 MCP,以便您的 AI 助手可以读取和写入 ProductPlan 数据。
什么是 ProductPlan?
ProductPlan 是 4000 多家产品团队使用的路线图软件。它处理路线图、OKR、想法发现和发布协调。
给开发者
productplan-mcp-server/
├── cmd/productplan/main.go # Entry point (~100 lines)
├── internal/
│ ├── api/ # ProductPlan API client
│ │ ├── client.go # HTTP client with caching, retry, rate limiting
│ │ ├── endpoints.go # 40+ API endpoint methods
│ │ └── formatters.go # Response enrichment for AI
│ ├── mcp/ # MCP protocol implementation
│ │ ├── server.go # JSON-RPC server, stdio I/O
│ │ ├── handler.go # Tool dispatch via registry
│ │ └── types.go # Protocol types
│ ├── tools/ # Tool definitions and handlers
│ │ ├── registry.go # Tool registration and dispatch
│ │ └── types.go # Typed argument structs for handlers
│ ├── cli/ # CLI commands (status, roadmaps, etc.)
│ │ └── cli.go
│ └── logging/ # Structured JSON logging
│ └── logger.go
├── pkg/productplan/ # Reusable utilities
│ ├── cache.go # LRU cache with TTL
│ ├── retry.go # Exponential backoff with jitter
│ ├── ratelimit.go # Adaptive rate limiting
│ ├── registry.go # ToolBuilder for schema generation
│ ├── requestid.go # Request tracing
│ └── errors.go # Error suggestions
└── evals/ # LLM evaluation test suite
├── tool_selection.json
├── confusion_pairs.json
└── argument_correctness.jsongit clone https://github.com/olgasafonova/productplan-mcp-server.git
cd productplan-mcp-server
go build -o productplan ./cmd/productplan为所有平台构建:
# macOS Apple Silicon
GOOS=darwin GOARCH=arm64 go build -o dist/productplan-darwin-arm64 ./cmd/productplan
# macOS Intel
GOOS=darwin GOARCH=amd64 go build -o dist/productplan-darwin-amd64 ./cmd/productplan
# Linux
GOOS=linux GOARCH=amd64 go build -o dist/productplan-linux-amd64 ./cmd/productplan
# Windows
GOOS=windows GOARCH=amd64 go build -o dist/productplan-windows-amd64.exe ./cmd/productplan运行所有测试:
go test ./...运行覆盖率:
go test ./... -cover运行基准测试:
go test ./internal/... -bench=. -benchmem运行评估套件:
./scripts/run-evals.sh覆盖率目标:
包 | 覆盖率 |
internal/mcp | 97% |
internal/logging | 97% |
internal/api | 95% |
internal/cli | 95% |
internal/tools | 90% |
提供 47 个工具:35 个读取工具和 12 个写入工具(基于操作):
读取工具:
路线图:
list_roadmaps,get_roadmap,get_roadmap_bars, `get_
Available Tools
50 toolsbulk_create_barsA
Create many bars on one roadmap in one call.
USE WHEN: "Add these 20 features to the roadmap", "Import this list as bars", "Create a bar per epic" For a single bar, use manage_bar instead. Each item needs name and a lane (lane name or lane_id), from the item or from set. Put shared values in set (e.g. set:{"lane":"Backend","legend":"Exploring"}). Up to 100 items. All items are validated against the roadmap before anything is written; one invalid item means nothing is sent. dry_run:true returns the exact POST payloads without writing. Parking follows manage_bar: a bar given starts_on and ends_on and no parked lands on the timeline (parked:false); an undated bar is parked by ProductPlan's default; a nested bar inherits its container's parked state. Writes run 4 at a time under the client's rate limiter. Returns per-item {index, bar_id, name, ok, error} and a summary like "Created 18 of 20 bars; 2 failed". The result is an error only when no bar was created. On a partial failure, retry only the failed items: resending the whole call would duplicate the bars that were created. FAILS WHEN: roadmap_id missing, items empty or over 100, an item without name or lane, a name not on the roadmap.
Inputs: items (required), roadmap_id (required), set, dry_run.
bulk_delete_barsADestructive
Permanently delete many bars in one call, verifying each delete.
USE WHEN: "Delete these bars", "Remove all the test bars I just made", "Clean up the parked duplicates" For a single bar, use manage_bar action=delete instead. Requires confirm:true. dry_run:true looks each bar up and returns its name without deleting, so the list can be checked first. Up to 100 bar_ids, each once. Each delete is verified by reading the bar back and expecting 404; a bar that still exists after a successful-looking delete is reported as failed. Deletes run 4 at a time under the client's rate limiter. Returns per-item {index, bar_id, ok, error} and a summary like "Deleted 9 of 10 bars; 1 failed". The result is an error only when no bar was deleted. FAILS WHEN: confirm is not true (and dry_run is not set), bar_ids empty, over 100, malformed, or repeated. WARNING: delete is permanent and cannot be undone.
Inputs: bar_ids (required), confirm, dry_run.
bulk_update_barsADestructive
Update many bars in one call: recolor, move lanes, retag, reschedule, set progress.
USE WHEN: "Color all these bars Committed", "Move these 12 bars to the Backend lane", "Set percent_done on every Q3 bar" For a single bar, use manage_bar instead. Put values shared by every bar in set (e.g. set:{"legend":"Committed"}) and per-bar values in items; an item's own fields override set. Up to 100 items; each bar_id may appear once. Every item is validated before anything is written: legend, lane, custom field labels, and dropdown values are checked against each bar's roadmap (one roadmap fetch per distinct roadmap; pass roadmap_id to skip looking up each bar's roadmap). One invalid item means nothing is sent, and the error lists every problem with the valid options. dry_run:true returns the exact PATCH payload per bar without writing. Writes run 4 at a time under the client's rate limiter. Returns per-item {index, bar_id, ok, error} and a summary like "Updated 28 of 30 bars; 2 failed". The result is an error only when no bar was updated. A partial failure is a normal result whose summary counts the failures; retry only the failed items. FAILS WHEN: items empty or over 100, a bar_id missing or repeated, a name not on the roadmap, legend_id or effort passed (see manage_bar).
Inputs: items (required), set, dry_run, roadmap_id.
check_statusARead-only
Check ProductPlan API status and authentication.
USE WHEN: "Is ProductPlan connected?", "Check API" For MCP server internals (cache stats, rate limits), use health_check instead.
Returns: data, summary
get_barARead-only
Get bar details including description, links, custom fields.
USE WHEN: "Tell me about this feature", "Bar details" Returns bar name, dates, description, links, custom fields, percent_done, and lane info. FAILS WHEN: bar_id not found (get valid IDs from get_roadmap_bars).
Inputs: bar_id (required). Returns: data, summary
get_bar_childrenARead-only
Get child bars nested under a parent bar.
USE WHEN: "Show sub-tasks", "Child items", "Break down this feature" FAILS WHEN: bar_id not found. Returns empty list if bar has no children (not all bars are containers).
Inputs: bar_id (required). Returns: data, summary
get_bar_commentsARead-only
Get comments on a bar.
USE WHEN: "Show comments", "What's the feedback on this bar?" For roadmap-level comments, use get_roadmap_comments instead. FAILS WHEN: bar_id not found.
Inputs: bar_id (required). Returns: data, summary
get_bar_connectionsARead-only
Get bar dependencies (what blocks what).
USE WHEN: "What depends on this?", "Show dependencies" FAILS WHEN: bar_id not found. Returns empty list if bar has no connections.
Inputs: bar_id (required). Returns: data, summary
get_bar_linksARead-only
Get external links on a bar (Jira, docs, designs).
USE WHEN: "What's linked?", "Show Jira tickets" FAILS WHEN: bar_id not found. Returns empty list if bar has no external links.
Inputs: bar_id (required). Returns: data, summary
get_ideaARead-only
Get idea details including description and metadata.
USE WHEN: "Tell me about this idea", "Full request details" FAILS WHEN: idea_id not found (get valid IDs from list_ideas).
Inputs: idea_id (required). Returns: data, summary
get_idea_formARead-only
Get idea form details with fields.
USE WHEN: "Show form fields", "What does this form collect?" FAILS WHEN: form_id not found (get valid IDs from list_idea_forms).
Inputs: form_id (required). Returns: data, summary
get_key_resultARead-only
Get key result details.
USE WHEN: "Tell me about this KR", "KR progress" FAILS WHEN: objective_id or key_result_id not found (use list_key_results to get valid KR IDs).
Inputs: objective_id (required), key_result_id (required). Returns: data, summary
get_launchARead-only
Get launch details with checklist.
USE WHEN: "Tell me about this launch", "Launch readiness" FAILS WHEN: launch_id not found (get valid IDs from list_launches).
Inputs: launch_id (required). Returns: data, summary
get_launch_sectionARead-only
Get a specific checklist section by ID.
USE WHEN: "Show one specific checklist section by ID" For all sections, use get_launch_sections. FAILS WHEN: launch_id or section_id not found.
Inputs: launch_id (required), section_id (required). Returns: data, summary
get_launch_sectionsARead-only
Get checklist sections for a launch.
USE WHEN: "Show sections", "Checklist categories" For one specific section, use get_launch_section. FAILS WHEN: launch_id not found.
Inputs: launch_id (required). Returns: data, summary
get_launch_taskARead-only
Get a specific launch task by ID.
USE WHEN: "Show one specific task by ID", "Check task assignment or status" For all tasks, use get_launch_tasks. FAILS WHEN: launch_id or task_id not found.
Inputs: task_id (required), launch_id (required). Returns: data, summary
get_launch_tasksARead-only
Get all tasks for a launch.
USE WHEN: "Show tasks", "What needs to be done?" For one specific task, use get_launch_task. FAILS WHEN: launch_id not found.
Inputs: launch_id (required). Returns: data, summary
get_objectiveARead-only
Get objective details with key results.
USE WHEN: "Tell me about objective X", "OKR progress" FAILS WHEN: objective_id not found (get valid IDs from list_objectives).
Inputs: objective_id (required). Returns: data, summary
get_opportunityARead-only
Get opportunity details with linked ideas.
USE WHEN: "Tell me about this opportunity" FAILS WHEN: opportunity_id not found (get valid IDs from list_opportunities).
Inputs: opportunity_id (required). Returns: data, summary
get_roadmapARead-only
Get roadmap settings and metadata.
USE WHEN: "Tell me about roadmap X", "Roadmap settings", "What custom fields does this roadmap have?" For all data in one call (bars, lanes, milestones), use get_roadmap_complete. Returns roadmap name, date range, sharing settings, and metadata, including the vocabulary bar writes must use: legends (names), lanes (names), custom_text_fields [{label}], and custom_dropdown_fields [{label, allowed_values}]. manage_bar and the bulk_*_bars tools validate against exactly these. FAILS WHEN: roadmap_id not found (get valid IDs from list_roadmaps first).
Inputs: roadmap_id (required). Returns: data, summary
get_roadmap_barsARead-only
Get bars (features/items) on a roadmap, optionally filtered.
USE WHEN: "What's on the roadmap?", "Show planned features", "What's starting in Q2?", "Bars in the Mobile lane tagged urgent" Server-side filters (sent to ProductPlan): name_contains, starts_after/starts_before, ends_after/ends_before (YYYY-MM-DD, inclusive), is_container, sort ("starts_on asc"). Client-side filters (applied here to every fetched bar, before the 50-item cap): lane (name or ID), legend (name), tag. Exact, case-insensitive. Returns bars with ID, name, starts_on/ends_on, lane_name and lane_id, legend, tags, percent_done, is_container, and parked. For a bar's description and custom fields, use get_bar. FAILS WHEN: roadmap_id not found (use list_roadmaps); a date is not YYYY-MM-DD; sort names a field outside the allowed list (the error lists them). Says "No bars matched the filters" when filters exclude everything.
Inputs: roadmap_id (required), tag, lane, sort, +7 more. Returns: data, summary
get_roadmap_commentsARead-only
Get roadmap-level comments (not bar comments).
USE WHEN: "Show roadmap comments", "Roadmap discussion" For bar-level comments, use get_bar_comments instead. Returns array of comments with author, body, and timestamp. FAILS WHEN: roadmap_id not found (use list_roadmaps).
Inputs: roadmap_id (required). Returns: data, summary
get_roadmap_completeARead-only
Get complete roadmap in one call. Details, bars, lanes, milestones combined.
USE WHEN: "Full roadmap overview", "Summarize roadmap X" For settings/metadata only, use get_roadmap. Returns combined roadmap details, bars, lanes, and milestones in one response. FAILS WHEN: roadmap_id not found (use list_roadmaps).
Inputs: roadmap_id (required). Returns: data, summary
get_roadmap_lanesARead-only
Get lanes (categories) on a roadmap. Lanes organize bars into rows.
USE WHEN: "What lanes are on the roadmap?", "Show categories" Returns array of lanes with ID, name, description, and position (the API exposes no lane color). FAILS WHEN: roadmap_id not found (use list_roadmaps).
Inputs: roadmap_id (required). Returns: data, summary
get_roadmap_legendsARead-only
Get the legend names (bar colors) for a roadmap.
USE WHEN: "What colors are available?", "Show the legend", "Which legend can I color bars with?"
Returns an array of legend names. The ProductPlan API exposes legends by name only: there is no legend ID or hex color to return. Pass a name as legend on manage_bar, bulk_update_bars, or bulk_create_bars to color a bar.
For custom field definitions (labels, dropdown allowed_values), use get_roadmap instead. To find bars by legend, use get_roadmap_bars with legend.
FAILS WHEN: roadmap_id not found (use list_roadmaps).
Inputs: roadmap_id (required). Returns: data, summary
get_roadmap_milestonesARead-only
Get milestones (key dates) on a roadmap.
USE WHEN: "What are the key dates?", "Show milestones" Returns array of milestones with ID, title, and date. FAILS WHEN: roadmap_id not found (use list_roadmaps).
Inputs: roadmap_id (required). Returns: data, summary
health_checkARead-only
Check MCP server health and cache stats.
USE WHEN: "Server status", "Rate limits", "Diagnose issues" For API connectivity only, use check_status instead. FAILS WHEN: deep=true and API is unreachable. Basic health (deep=false) always succeeds if server is running.
Inputs: deep. Returns: data, summary
list_all_customersARead-only
List all customers across ideas. Returns customer names and linked idea counts.
USE WHEN: "Who are our customers?", "All feedback sources" FAILS WHEN: API token invalid.
Returns: data, summary
list_all_tagsARead-only
List all tags used across ideas. Returns tag names and usage counts across ideas.
USE WHEN: "What tags exist?", "Show categories" FAILS WHEN: API token invalid.
Returns: data, summary
list_idea_formsARead-only
List idea submission forms.
USE WHEN: "Show feedback forms", "What forms exist?" Returns array of forms with ID and name. FAILS WHEN: API token invalid.
Returns: data, summary
list_ideasARead-only
List all ideas in discovery pipeline. START HERE for ideas.
USE WHEN: "Show customer feedback", "What ideas do we have?", "Ideas mentioning SSO" Optional server-side filters: name_contains (case-insensitive), channel (exact); sort ("name asc"). Returns ideas with ID, name, channel, and opportunities_count. FAILS WHEN: API token invalid; sort names a field outside the allowed list (the error lists them). Returns empty list if no ideas exist.
Inputs: sort, channel, name_contains. Returns: data, summary
list_key_resultsARead-only
List key results for an objective.
USE WHEN: "What are the KRs?", "Show metrics" Returns array of key results with ID, name, target_value, and current_value. FAILS WHEN: objective_id not found (use list_objectives).
Inputs: objective_id (required). Returns: data, summary
list_launchesARead-only
List all launches. START HERE for launches.
USE WHEN: "Show launches", "Release schedule", "Launches this quarter" Optional server-side filters: name_contains (case-insensitive), status (exact), launch_after/launch_before (YYYY-MM-DD, inclusive); sort ("launch_date asc"). Returns array of launches with ID, name, launch_date, status, and progress. FAILS WHEN: API token invalid; a date is not YYYY-MM-DD; sort names a field outside the allowed list (the error lists them). Returns empty list if no launches exist.
Inputs: sort, status, launch_after, launch_before, +1 more. Returns: data, summary
list_objectivesARead-only
List all OKR objectives. START HERE for OKRs.
USE WHEN: "Show OKRs", "What are our objectives?" Returns array of objectives with ID, name, risk_status, start_date, end_date, and key_results_count. FAILS WHEN: API token invalid. Returns empty list if no objectives exist.
Returns: data, summary
list_opportunitiesARead-only
List all opportunities. START HERE for discovery.
USE WHEN: "Show opportunities", "Discovery pipeline", "Opportunities about onboarding" Optional server-side filters: problem_contains (case-insensitive), workflow_status (exact); sort ("ideas_count desc"). Returns array of opportunities with ID, problem_statement, workflow_status, and linked idea count. FAILS WHEN: API token invalid; sort names a field outside the allowed list (the error lists them). Returns empty list if no opportunities exist.
Inputs: sort, workflow_status, problem_contains. Returns: data, summary
list_roadmapsARead-only
List all roadmaps. START HERE to get roadmap IDs.
USE WHEN: "Show my roadmaps", "What roadmaps do I have?", "Find the Mobile roadmap" Optional server-side filters: name_contains (case-insensitive); sort ("name asc", "updated_at desc"). Returns roadmaps with ID, name, and updated_at. FAILS WHEN: API token invalid or expired (check PRODUCTPLAN_API_TOKEN env var); sort names a field outside the allowed list (the error lists them).
Inputs: sort, name_contains. Returns: data, summary
list_teamsARead-only
List all teams in account.
USE WHEN: "What teams exist?", "Team structure" For individual user details, use list_users instead.
Returns: data, summary
list_usersARead-only
List all users in account.
USE WHEN: "Who has access?", "Team members" Returns array of users with ID, name, email, and role. Use user IDs from this tool when assigning launch tasks via manage_launch_task.
Returns: data, summary
manage_barADestructive
Create, update, or delete a bar on a roadmap.
USE WHEN: "Add feature", "Update dates", "Delete item", "Change color", "Nest this bar under that one" For many bars at once (e.g. "color these 30 bars"), use bulk_update_bars, bulk_create_bars, or bulk_delete_bars instead. Actions: create (roadmap_id + name + lane or lane_id), update (bar_id + any fields), delete (bar_id) Color a bar with legend (a legend NAME from get_roadmap_legends); legend:"" or clear_legend:true removes the color. Omitted fields are never sent, so an update only touches what you pass. New bars are parked (off the timeline) by ProductPlan's default. When you give starts_on and ends_on on create and leave parked unset, the bar defaults to parked:false so it lands on the timeline; a nested bar inherits its container's parked state. Names (legend, lane, custom field labels, dropdown values) are checked against the roadmap before writing, matched case-insensitively, and sent in canonical spelling. Returns: create gives the new bar's id plus the bar read back; update gives the exact fields sent. FAILS WHEN: create without roadmap_id, name, or a lane; update/delete without bar_id; a legend, lane, or custom field name is not on the roadmap (the error lists the valid ones); container_bar_id on a bar without both dates, or a parked state that differs from the container's. legend_id and effort are rejected with an explanation (legend_id would wipe the bar's color). A bar cannot be un-nested via the API once container_bar_id is set. WARNING: delete is permanent and cannot be undone.
Inputs: action (required), lane, name, tags, +20 more.
manage_bar_connectionADestructive
Create or delete dependency between bars.
USE WHEN: "Link features", "Add dependency", "Remove dependency" Actions: create (target_bar_id), delete (connection_id) FAILS WHEN: create without target_bar_id, delete without connection_id (get IDs from get_bar_connections).
Inputs: action (required), bar_id (required), connection_id, target_bar_id.
manage_bar_linkADestructive
Create or delete external link on a bar.
USE WHEN: "Link Jira ticket", "Add design doc", "Remove link" Actions: create (url), delete (link_id) FAILS WHEN: create without url, delete without link_id (get IDs from get_bar_links). Note: update not available via API; delete and re-create instead.
Inputs: action (required), bar_id (required), url, name, +1 more.
manage_ideaADestructive
Create or update an idea. Note: delete not available via API.
USE WHEN: "Add idea", "Update idea status" Actions: create (title), update (idea_id) Returns the created/updated idea object. FAILS WHEN: create without title, update without idea_id. Note: delete is not available via the ProductPlan API; archive ideas by updating status instead.
Inputs: action (required), title, status, idea_id, +1 more.
manage_key_resultADestructive
Create, update, or delete a key result.
USE WHEN: "Add KR", "Update progress", "Delete KR" Actions: create (name+target), update (key_result_id), delete (key_result_id) Returns the created/updated key result, or confirmation on delete. FAILS WHEN: create without name or target_value, update/delete without key_result_id (use list_key_results).
Inputs: action (required), objective_id (required), name, target_value, +2 more.
manage_laneADestructive
Create, update, or delete a lane on a roadmap.
USE WHEN: "Add Backend lane", "Rename Mobile lane", "Move lane to the top", "Delete lane" Actions: create (name; optional description, position), update (lane_id plus any of name, description, position), delete (lane_id) Returns the created/updated lane object, or confirmation on delete. FAILS WHEN: create without name, update/delete without lane_id (get IDs from get_roadmap_lanes), color passed (lanes have no settable color in the API). WARNING: delete removes the lane and unassigns all bars in it.
Inputs: action (required), roadmap_id (required), name, color, +3 more.
manage_launchADestructive
Create, update, or delete a launch.
USE WHEN: "Create launch", "Update date", "Delete launch" Actions: create (name+date), update (launch_id), delete (launch_id) Returns the created/updated launch object, or confirmation on delete. FAILS WHEN: create without name or date, update/delete without launch_id, date not in YYYY-MM-DD format. WARNING: delete removes the launch and all its sections and tasks.
Inputs: action (required), date, name, launch_id, +1 more.
manage_launch_sectionADestructive
Create, update, or delete a checklist section.
USE WHEN: "Add Marketing section", "Rename section", "Delete section" Actions: create (name), update (section_id), delete (section_id) Returns the created/updated section object, or confirmation on delete. FAILS WHEN: create without name, update/delete without section_id (get IDs from get_launch_sections). WARNING: delete removes the section and all tasks in it.
Inputs: action (required), launch_id (required), name, section_id.
manage_launch_taskADestructive
Create, update, or delete a launch task.
USE WHEN: "Add task", "Mark complete", "Assign task", "Delete task" Actions: create (name+section_id), update (task_id), delete (task_id) Returns the created/updated task object, or confirmation on delete. FAILS WHEN: create without name or section_id, update/delete without task_id (get IDs from get_launch_tasks). Use list_users to get valid assigned_user_id values.
Inputs: action (required), launch_id (required), name, status, +5 more.
manage_milestoneADestructive
Create, update, or delete a milestone on a roadmap.
USE WHEN: "Add launch milestone", "Move demo date", "Delete milestone" Actions: create (title+date), update (milestone_id), delete (milestone_id) Returns the created/updated milestone object, or confirmation on delete. FAILS WHEN: create without title or date, update/delete without milestone_id (get IDs from get_roadmap_milestones), date not in YYYY-MM-DD format.
Inputs: action (required), roadmap_id (required), date, title, +1 more.
manage_objectiveADestructive
Create, update, or delete an objective.
USE WHEN: "Add Q1 objective", "Update objective", "Delete OKR" Actions: create (name), update (objective_id), delete (objective_id) Returns the created/updated objective, or confirmation on delete. FAILS WHEN: create without name, update/delete without objective_id. WARNING: delete also removes all key results under this objective.
Inputs: action (required), name, time_frame, description, +1 more.
manage_opportunityADestructive
Create or update an opportunity. Note: delete not available via API.
USE WHEN: "Create opportunity", "Update problem" Actions: create (problem_statement), update (opportunity_id) Returns the created/updated opportunity object. FAILS WHEN: create without problem_statement, update without opportunity_id (get IDs from list_opportunities). Note: delete is not available via the ProductPlan API; archive opportunities by updating workflow_status instead.
Inputs: action (required), description, opportunity_id, workflow_status, +1 more.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 50 tools
Each tool pairs a distinct resource with a clear read/write action, and the get_/list_/manage_/bulk_ prefixes make intent obvious. Plural-versus-singular pairs like get_launch_sections and get_launch_section are explicitly annotated so an agent can reliably choose the right one.
The set follows a consistent convention: list_* for collections, get_* for single objects/details, manage_* for create/update/delete, and bulk_*_bars for multi-bar operations. Minor exceptions such as check_status and health_check are still self-describing and do not break the overall pattern.
Fifty tools is far beyond the typical well-scoped MCP surface and forces agents to consider a very long list for any task. The tools are individually justified, but the sheer count makes the server feel heavy and increases selection cost.
The server covers nearly every lifecycle needed for ProductPlan work: bars, lanes, milestones, launches, sections, tasks, objectives, key results, ideas, opportunities, links, and dependencies all have appropriate read/write tools. The main gaps are that roadmaps themselves cannot be created/updated/deleted and comments are read-only, though these are likely API constraints rather than design oversights.
Maintenance
Related MCP Connectors
MCP server that lets AI assistants use all OneSchema features exposed via the public API.
MCP server for AI access to SmartBear tools, including BugSnag, Reflect, Swagger, PactFlow, QTM4J.
MCP server for generating rough-draft project plans from natural-language prompts.
Official Fibery MCP server: query and update your workspace from any AI client via OAuth.
Related MCP Servers
- AlicenseAqualityBmaintenanceMCP server for controlling Miro whiteboards with AI assistants. Create and manage stickies, shapes, frames, connectors, and more.8832MIT
- AlicenseCqualityDmaintenanceMCP server for Plane integration, enabling AI agents to interact with Plane APIs and services for project management tasks.100MIT
- AlicenseAqualityDmaintenanceMCP server exposing 40 best-practice product management skills plus workflows and lifecycle tools, instantly accessible to any AI via Model Context Protocol.19152 npm21Apache 2.0
- FlicenseNot gradedqualityDmaintenanceAn unofficial MCP server for DevRev integration, enabling AI assistants to search, manage works, explore product hierarchies, and retrieve object details seamlessly.3-