Roadmap
Enables AI orchestrators to manage hierarchical implementation roadmaps with phases, tasks, and plan-change tracking.
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 / canceledstatuses. - "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 (
aioruser) 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-postgresplugin 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 thewithvibe_pluginsdatabase and injectsDATABASE_URL+PGSCHEMAat spawn. The plugin role can't reach the mainwithvibeDB. - State surface: four tables in db.js —
plan(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
- Workspace admin → Plugins → Install plugin.
- Paste the contents of manifest.yaml into the editor.
- The platform pulls / locates the image and registers the plugin.
- 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.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.