Roadmap

Roadmap

Enables AI orchestrators to manage hierarchical implementation roadmaps with phases, tasks, and plan-change tracking.

Category
访问服务器

README

WithVibe Roadmap Plugin

A per-env implementation roadmap board for WithVibe — Postgres-backed, with an MCP server the AI orchestrator drives directly.

Multi-phase features drift. Plans change mid-flight. After a few sessions you lose track of what shipped, what's pending, and which step you're actually on. This plugin gives the AI a structured place to keep that state — and the user a live view of it.

Feature: Auth rewrite                          ▓▓▓▓░░░░░░ 4/11 tasks
you are here →  API layer  ›  Wire DTO validation

[x] Phase 1: Schema (done)
[~] Phase 2: API layer (in progress)
    [x] Endpoint scaffolding
    [x] Auth guard
    [~] Wire DTO validation     ← active
    [ ] Error envelope
[ ] Phase 3: Web
[ ] Phase 4: QA

— Plan changes —
2026-05-28  AI added Phase 4 (QA) — reason: integration tests not in original scope
2026-05-29  user removed "Add caching" — reason: deferred to follow-up

What's in the box

  • Hierarchical roadmap — feature → phase → task, with pending / in_progress / done / blocked / deferred / canceled statuses.
  • "You are here" marker — exactly one task is active at a time; the banner is the recovery anchor when you context-switch back.
  • Plan-change log — every add/remove/rename/status-change records an actor (ai or user) and an optional reason. The trail makes drift visible.
  • MCP tools the AI orchestrator uses without prompting: propose_plan, set_active_task, complete_task, add_note, update_task, log_change, …
  • Live UI — htmx-based, polls a tiny version endpoint so the iframe updates within ~4s of any AI change without scroll jumps.
  • Persistent — uses WithVibe's shared-postgres plugin storage, so state survives container restarts.

Architecture

manifest.yaml ──→ WithVibe spawns one container per env
       ↓
   server.js (express)
   ├── /health         platform health probe
   ├── /mcp            AI orchestrator's MCP endpoint  ─→ mcp.js
   ├── /ui             dark-themed htmx UI             ─→ ui.js
   └── /ui/version     cheap freshness check (drives live updates)
                ↓
              db.js (pg)
                ↓
          shared-postgres
          per-env schema
          plan / phase / task / plan_event
  • Scope: env — one container per (env, plugin). Each env gets its own roadmap, its own Postgres schema, its own port.
  • Storage: shared-postgres — the platform provisions a dedicated role + schema in the withvibe_plugins database and injects DATABASE_URL + PGSCHEMA at spawn. The plugin role can't reach the main withvibe DB.
  • State surface: four tables in db.jsplan (singleton), phase, task, plan_event (change log).

MCP tools

All tools return the full updated plan in the trailing text so the AI never holds stale state.

Tool Purpose
get_plan Read the full roadmap + active marker + last 20 events.
propose_plan Replace the whole roadmap (initial planning or a major re-plan).
add_phase / add_task Append work mid-flight; reason recorded.
set_active_task Mark the "you are here" anchor before starting work.
complete_task Mark done with a 1-line outcome; auto-closes the phase if every task is done.
update_task / update_phase Edit title/description/status (blocked, deferred, canceled) with reason.
add_note Append a markdown note — decisions, things tried, references.
log_change Narrate a plan-level decision without mutating data.

See mcp.js for the full schemas.

Build

docker build -t local/roadmap:2.0 .

The image is multi-arch-friendly via Node 20 Alpine, ~150 MB, no native build steps.

Install in WithVibe

  1. Workspace admin → PluginsInstall plugin.
  2. Paste the contents of manifest.yaml into the editor.
  3. The platform pulls / locates the image and registers the plugin.
  4. Open any env → the Roadmap tab appears in the plugin panel.

To update after rebuilding the image, hit Update on the plugin row in the admin list — running instances are stopped so the next env start picks up the new image.

Manifest

The manifest is the entire install input. The fields that matter for this plugin:

id: withvibe.roadmap          # URL + tool prefix; reverse-DNS-ish
name: Roadmap                 # display name in the env's plugin panel
version: 2.0.0
icon: list-todo               # Lucide icon
image: local/roadmap:2.0    # OCI ref

scope: env                    # one container per env
storage:
  kind: shared-postgres       # platform-managed Postgres role + schema

ui:
  path: /ui                   # iframe entry point
  websocket: false

mcp:
  enabled: true
  path: /mcp                  # MCP endpoint the AI auto-discovers

Local development

# Build
docker build -t local/roadmap:2.0 .

# Run standalone (requires a Postgres reachable via DATABASE_URL)
docker run --rm -p 8080:8080 \
  -e DATABASE_URL="postgres://user:pass@host.docker.internal:5432/withvibe_plugins" \
  -e PGSCHEMA="roadmap_dev" \
  local/roadmap:2.0

# Then open http://localhost:8080/ui

When developing against a real WithVibe install, hit Update in the admin Plugins page after each rebuild — that stops running instances so the next env start picks up the new image.

Repository layout

manifest.yaml   WithVibe plugin manifest
Dockerfile      builds the runtime image
package.json    npm deps (express, pg, @modelcontextprotocol/sdk, zod)
server.js       express entry, HTTP + MCP routes
db.js           pg pool, schema init, query helpers
mcp.js          MCP tool registrations
ui.js           htmx UI rendering + dark-themed CSS

Contributing

Issues and PRs welcome. Keep changes focused and document the why in commits.

License

Apache License 2.0. See NOTICE for attribution.

推荐服务器

Baidu Map

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。

官方
精选
JavaScript
Playwright MCP Server

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。

官方
精选
TypeScript
Magic Component Platform (MCP)

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。

官方
精选
本地
TypeScript
Audiense Insights MCP Server

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。

官方
精选
本地
TypeScript
VeyraX

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。

官方
精选
本地
graphlit-mcp-server

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。

官方
精选
TypeScript
Kagi MCP Server

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。

官方
精选
Python
e2b-mcp-server

e2b-mcp-server

使用 MCP 通过 e2b 运行代码。

官方
精选
Neon MCP Server

Neon MCP Server

用于与 Neon 管理 API 和数据库交互的 MCP 服务器

官方
精选
Exa MCP Server

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。

官方
精选