MCP 服务器

pnCore

Enables software delivery as a sequence of named, gated workflow steps, from discovery and planning through build, review, and handoff, with verifiable gates and resumable state outside the chat.

README

pnCore — v0.18.7

pnCore is an MCP server and Cursor plugin. It runs software delivery as a sequence of named, gated steps instead of one long chat.

It ships discovery, planning, skeptic challenge, design, audits, assets, and delivery through a deterministic workflow_step engine — backed by skills, agents, rules, and pn-core:// resources, not a folder of prompts.

Catalog: 170 skills, 9 public agents + 6 internal orchestration agents, 31 visible slash palette files (30 under pn submenu + /pn stub) + 18 palette-hidden surgical commands (49 command files total), 26 MCP tools, 16 workflow types, plus pn-core:// resources and prompts.

Why this exists

A single chat can plan a feature, write the code, and tell you it's done — but there's no checkpoint in between. If it goes sideways on step three, you're rereading the whole transcript to find out where. Asking it to redo one part usually means starting the conversation over, and whatever context got it that far is gone.

pnCore moves that step list out of the chat. workflow_step(type, index, state) runs one step at a time — discovery, plan, build, review — and hands back the next instruction. A skeptic pass, a human gate, or workflow_verify decides whether a step actually passed, not the model's own say-so. If you get disconnected, workflow_state_save and workflow_state_load pick the run back up where it left off.

What it is

Deterministic. workflow_step(type, index, state) decides the next instruction. Steps aren't skipped by assumption — skipping one is a gate decision, logged in state.

Gated. Skeptic, human, and workflow_verify gates are built into the engine. Intent is full auto, design focused, or involved — involved means you approve discovery, plan, specialists, and review before they run.

Multi-surface. One canonical tree in packages/pn-core-mcp/content/. The MCP runs the engine in any client. The Cursor plugin adds the / palette, file-glob rules, agent selector, and stop hook. Pi registers the same 26 tools natively.

Resumable. Every run has a run_id. Handoff lines and usage land in JSONL. After a disconnect, load state and continue the same step list.

Install

Prerequisite: Node.js 22+.

Call health first — it should return version, calendarDateUtc, and capabilities. If it doesn't, nothing downstream (skills, agents, workflow_step) will load either.

Cursor — MCP (one-click)

Or add manually to ~/.cursor/mcp.json:

{
  "mcpServers": {
    "pn-core": {
      "command": "npx",
      "args": ["-y", "--package=git+https://github.com/perniemann/pnCore.git#main", "--", "pn-core"],
      "env": { "GIT_TERMINAL_PROMPT": "0", "GIT_ASKPASS": "echo" }
    }
  }
}

GIT_TERMINAL_PROMPT=0 and GIT_ASKPASS=echo fail fast when git cannot authenticate (missing credentials, rate limit, or a bad URL). Without them, a credential prompt on piped stdin leaves Cursor in a forever loading state.

Cursor — Plugin

From your target project directory:

npx github:perniemann/pnCore plugin-install

Copies commands, rules, skills, agents, config, and hooks into .cursor/ and .cursor-plugin/. Reload Cursor, then run /pn-new.

MCP or plugin? MCP is the engine. The plugin is the Cursor surface (slash palette, rules, agents, stop hook). The intended Cursor setup is both. Details: Plugin vs MCP.

pi.dev (Pi coding agent)

pi install git:github.com/perniemann/pnCore@main

Pi surfaces a single /pn entry. Direct invoke: /pn pn-build. Native tools ship with pi install git:…/pnCore or pi install . from this repo root after npm run build:mcp. Cursor and Claude Code still use the stdio MCP server. See ADR-0008 and ADR-0009.

Claude Code

Use the same MCP JSON. Settings → MCP Servers, paste the block, enable it. See Anthropic's MCP documentation.

Windows, Cloud Agents, this-checkout node paths, and first-npx timeouts: packages/pn-core-mcp/README.md. Clone-and-develop: CONTRIBUTING.md.

Quick start

Install MCP and/or plugin (above).
Call health — version, UTC date, capabilities.
New repo or greenfield: /pn-new or workflow_step("project_kickoff", 0, {}) — discovery, refs, PRD, and design docs under docs/refs/.
Build or extend: /pn-build or workflow_step("full_dev", 0, {}) — skips kickoff when the project already has context.
Before ship: /pn-deliver or /pn-frontend-audit.

Copy-paste first message:

pn-new ▲

Build [your-project-name] — [one-line description].
References: [path or "in .ref/"] (pitch, requirements, design assets).
Analyze both: prior art and design.
Intent: Involved — full gates at discovery, plan, specialists, and review.
Delivery tier: full. Design ambition: distinctive.

Answer Step 0: Yes, Both. Step 1: (3) Involved.

More prompts: docs/how-to-use-guide.md.

Workflows

Call list_workflow_types for live step counts.

Use it for	Workflow	Cursor	MCP entry
New project: refs, discovery, PRD, design	`project_kickoff`	`/pn-new` → Involved	`workflow_step("project_kickoff", 0, {})`
Build a feature or product end-to-end	`full_dev`	`/pn-build`	`workflow_step("full_dev", 0, {})`
Design-first UI	`design`	`/pn-design`	`workflow_step("design", 0, {})`
Scored frontend diagnosis + fix roadmap	`frontend_audit`	`/pn-frontend-audit`	`workflow_step("frontend_audit", 0, {})`
API, security, data, errors, performance	`backend_audit`	`/pn-backend-audit`	`workflow_step("backend_audit", 0, {})`
SVG, raster, or placeholders	`svg_create` / `image_create`	`/pn-assets`	`workflow_step("svg_create" \| "image_create", 0, {})`
Evidence-led strategy brief	`business_strategy`	`/pn-strategy`	`workflow_step("business_strategy", 0, {})`
Competing implementations (2–3 worktrees)	`implementation_tournament`	`/pn-best-of-n` (`bestOfN.enabled: true`)	`workflow_step("implementation_tournament", 0, {})`
Multi-slice hierarchical build	`feature_program`	`/pn-program` (preview; `featureProgram: true`)	`workflow_step("feature_program", 0, {})`

Also on the engine: visual_tweak, prompt_optimize, game_feature, engine_feature (Unreal / Godot), fsi_analyst_draft, media_director. Full table and aliases: list_workflow_types.

Example: `full_dev`

#	Who	Action
1	You	`/pn-new` or "Start a new project with full dev workflow." Choose full auto, design focused, or involved.
2	Agent	`workflow_step("full_dev", 0)` — discovery: purpose, users, stack, references, existing code.
3	Agent	Prior-art and research pass for your stack.
4	Agent	`workflow_step("full_dev", 2)` — roadmap, phases, architecture, plan under `docs/plans/`.
5	You	Review plan. Involved intent runs skeptic challenge; approve before build.
6	Agent	`workflow_step("full_dev", 3)` — specialist routing (frontend, backend, testing, …).
7	Agent	`workflow_step("full_dev", 4)` — specialists build; UI assets created when in scope.
8	Agent	Optional merge phase (`mergePhaseFullDev`): reconcile parallel work, verify build.
9	Agent	`workflow_step("full_dev", 5)` — review + optimize against the plan.
10	You	`/pn-deliver` for handoff pack, or `/pn-frontend-audit` for a scored quality gate.

Tool steps are 0-based. Resume after disconnect: workflow_state_save then workflow_state_load. Schema: pn-core://reference/workflow-state-schema.md.

Design-first: workflow_step("design", 0) instead of full_dev. Load .pncore-design.md via /pn-setup.

Game / 3D: workflow_step("game_feature", 0) for feature loops. For full builds, use full_dev and name the stack in discovery.

What changes

Without pnCore	With pnCore
One long transcript	Named steps, each with a `run_id`
"Done" means the model stopped talking	Skeptic, `workflow_verify`, or a delivery pack decide
A retry starts a new chat	`workflow_state_load` picks up where it stopped
Skills live in a folder you have to remember to load	The engine loads gates and skills automatically
Works in Cursor chat, nowhere else	MCP runs in any client; Pi has native tools

Three tier concepts (do not conflate them): delivery tier (MVP/Full), context tier (1–4 reading depth), model tier (fast / standard / premium / premium_thinking / long_horizon). Loop orchestration: suggest_model_tier with role: orchestrator → long_horizon. See pn-core://reference/delivery-tier-criteria.md and MCP tools.

Load before a build session: pn-core://reference/best-practices.md, pn-core://reference/loop-orchestration-guide.md, pn-core://reference/aesthetics-baseline.md, and health for the current UTC date.

Honest edges

Situation	What happens	What to do
A one-line typo fix	The workflow overhead isn't worth it	Ask directly, skip `workflow_step`
Plugin without MCP	Slash templates only — no `workflow_step` engine	Install MCP for Cursor, or both
First npx on a git+https package URL	Cursor MCP can time out on a cold clone	Pre-warm once, then reload. Matrix: MCP README
`feature_program` / `bestOfN`	Preview flags; off by default	Set `featureProgram: true` or `bestOfN.enabled: true`
Vue, Svelte, Angular, Unity	Limited support	Prefer React, Astro, Next, vanilla web, Node, Three.js / Babylon, n8n, web3

Best fit: teams building with Cursor on React, Astro, Next.js, vanilla web, Node backends, Three.js / Babylon / gamedev, n8n, and web3. Inventory: docs/plugin-reference.md.

Documentation

Guide	What's in it
docs/how-to-use-guide.md	Copy-paste prompts, example flows, MCP-only bootstrap
docs/mcp-usage-guide.md	MCP tools, resources, workflow patterns, state/handoff
docs/plugin-reference.md	Rules, skills, agents, commands, hooks
packages/pn-core-mcp/README.md	MCP config, 26 tools, env vars, error codes, resources
docs/companion-mcp-catalog.md	Companion MCPs (Octocode, Stripe, n8n, …)
docs/pitch-to-app-example.md	End-to-end pitch-to-app walkthrough
packages/pn-core-mcp/content/docs/starting-new-project.md	Kickoff and `docs/refs/` setup
CONTRIBUTING.md	Workspace map, scripts, PR workflow, ADR policy
plugins/pnCore/CHANGELOG.md	Release history

Scripts

Contributor scripts, local MCP config, and how to develop pnCore live in CONTRIBUTING.md. Repo layout: docs/folder-structure.md.

License

MIT — see LICENSE.