@wairon/cli 5.1.1-dev.100 → 5.1.1-dev.102

package/README.md

@@ -3,22 +3,25 @@
 **Waffle AIron** — an AI-Driven Development (AIDD) support tool built around
 **Spec-Driven Development (SDD)**.
 
-> **Status:** stable, actively developed — the SDD spec tree, the conformance gate (a 40-plus-rule
-> registry with narrative control-flow validation, technology
-> boundaries, and per-spec `lint.allow`), spec-derived agent topology,
-> skills, the MCP server, diagrams (Mermaid, interactive canvas + ERD,
-> draw.io/Excalidraw), extension packs (injectable profiles, language
-> tables, and rules), and a self-hostable HTTP **hosting server**
-> (`wairon serve` — HTTP MCP for many isolated projects) are working. See the
-> [roadmap](docs/roadmap.md) for what's next.
+> **Status:** stable, actively developed — the SDD spec tree, the conformance gate (a
+> documented rule registry — `wairon rules list` — with narrative control-flow
+> validation, code↔spec conformance, technology boundaries, and per-spec
+> `lint.allow`), human approval (`wairon lock`) with a CI merge gate
+> (`wairon lock-check`), members and cross-project references, live agent
+> briefs and skills, the MCP server, the design export, diagrams (Mermaid,
+> interactive canvas + ERD, draw.io/Excalidraw), extension packs (injectable
+> profiles, language tables, and rules), and a self-hostable HTTP **hosting
+> server** (`wairon serve` — HTTP MCP for many isolated projects, with a web UI)
+> are working. See the [roadmap](docs/roadmap.md) for what's next.
 
 ---
 
 ## What is it?
 
 `wairon` turns a validated **spec tree** into (a) an enforced architecture
-**conformance gate** and (b) a ready-made **agent topology + skills** that any AI
-coding tool (Claude Code, Gemini CLI, …) consumes to do the work itself.
+**conformance gate**, (b) a recorded **approval** that CI can check, and (c) a
+ready-made **agent topology + skills** that any AI coding tool (Claude Code,
+Gemini CLI, …) consumes to do the work itself.
 
 It is optional and additive: if a project has a `.wai/specs/` tree the workflow
 is active; otherwise you ignore wairon and work normally. wairon does **not** run
@@ -27,17 +30,22 @@ or orchestrate AI sessions — it *equips* the session you already use.
 ```
 .wai/specs/  (the SDD spec tree — source of truth)
    │
-   ├── wairon validate ──▶ architecture-conformance gate
-   │                       (stereotype dependency rules, contract↔impl
-   │                        symmetry, reference integrity, cycle detection)
+   ├── wairon validate ────▶ architecture-conformance gate
+   │                          (stereotype dependency rules, contract↔impl
+   │                           symmetry, reference integrity, cycle detection)
    │
-   └── wairon generate ──▶ .claude/agents/  .gemini/agents/  (derived subagents)
-                           + SDD skills installed into each tool
-                           + wairon MCP server (sdd_* tools)
+   ├── wairon lock ────────▶ .wai/lock.json  (the human approval, committed)
+   │     └ wairon lock-check ▶ CI merge gate: is this the approved design?
+   │
+   └── wairon MCP server ──▶ sdd_* tools + live agent briefs (sdd_get_agent_brief)
+                              + SDD skills installed into each tool
 ```
 
-The host AI tool spawns the generated agents as its **own native subagents**,
-guided by the SDD skills and the `sdd_*` MCP tools.
+The host AI tool spawns subagents of its own from the **live briefs** — each
+composed from the current spec tree on every call — guided by the SDD skills
+and the `sdd_*` MCP tools. Agent files on disk (`.claude/agents/`,
+`.gemini/agents/`) are an optional materialized view of the same topology:
+off by default, on with `rules.materializeAgentFiles: true`.
 
 ---
 
@@ -49,7 +57,7 @@ A complete system is specified top-down across six levels:
 |-------|------|-----------------|
 | **L0 System** | `.wai/specs/.index.yaml` | Vision, boundaries, global requirements, target language |
 | **L1 Subsystem** | `.wai/specs/<sub>/.index.yaml` | An isolated software service, its public interfaces, and trusted links |
-| **L2 Component** | `…/<component>/.index.yaml` | A building block (Portal, Orchestrator, Supervisor, Actor, Store, Index, Registry, Adapter, Observer, Specialist) or pattern (Repository, Gateway) + `owns`/`dependsOn` |
+| **L2 Component** | `…/<component>/.index.yaml` | A building block (Portal, Orchestrator, Supervisor, Actor, Store, Index, Query, Registry, Adapter, Observer, and View for UIs) or pattern (Repository; FeatureComponent and RouterComponent for UIs) + `owns`/`dependsOn`. A gateway is a Portal with the `gateway` variant; logic is an Orchestrator whose `dependencyClass` bounds what it may depend on |
 | **L3 Interface** | `…/<component>/.interface.yaml` | Method signatures & structured params (+ optional wire endpoint bindings) |
 | **L4 Implementation** | `…/<component>/.implementation.yaml` | Concrete implementation of a contract: narrative detail level, bound `technologies` (the swap seam for vendors/engines), optional source path |
 | **L5 Narrative** | (within L4) | Step-by-step method logic as a flat numbered list — `call` steps resolve to real dependency methods, and flow steps (`branch`/`switch`/`loop`/`try`/`jump`/`return`/`throw`) jump by step number |
@@ -76,7 +84,13 @@ registry:
 
 - a **`system-architect`** from L0,
 - a **`<subsystem>-owner`** per subsystem,
-- a **`<component>-implementer`** per component.
+- a **`<component>-implementer`** per component, when
+  `rules.generateComponentImplementers` is on.
+
+Each agent is served as a **live brief** (`sdd_get_agent_brief`, or
+`wairon agent brief <id>`), composed from the current tree whenever it is
+asked for, so it is never stale. Writing them to disk as agent files is opt-in
+(`rules.materializeAgentFiles: true`).
 
 A **domain** is a unit of ownership. Subsystems yield spec-backed domains
 automatically; you can also declare **free-standing domains** (docs, infra,
@@ -102,8 +116,19 @@ irm https://raw.githubusercontent.com/SYW-Apps/Waffle-AIron/main/install.ps1 | i
 curl -fsSL https://raw.githubusercontent.com/SYW-Apps/Waffle-AIron/main/install.sh | sh
 ```
 
-Both `wairon` and `wai` are registered as commands. (wairon is distributed
-as standalone binaries — it is not published to npm.)
+Both `wairon` and `wai` are registered as commands.
+
+### npm
+
+wairon is also published to npm as `@wairon/cli` (Node 18+):
+
+```sh
+npm install --save-dev @wairon/cli    # per project; run it with npx wairon
+npm install --global @wairon/cli      # or machine-wide
+```
+
+The dist-tags are the release channels: `latest` (stable), `beta`, `preview`
+and `dev`, e.g. `npm install -D @wairon/cli@dev`.
 
 ### Local development
 
@@ -138,15 +163,28 @@ merge to `dev`. Installing from npm, the channels are the dist-tags:
 
 ```sh
 cd my-project
-wairon init                 # bootstrap .wai/ (spec tree, context, skills, MCP-ready)
+wairon init                 # bootstrap .wai/ (spec tree, context, skills, MCP registration)
+
+# 1. Design: in your AI tool, use the sdd-architect skill, which authors the
+#    spec tree through the sdd_* MCP tools.
+
+wairon status               # the tree, its authoring readiness, and its approval state
+wairon validate             # architecture-conformance gate (--ci in CI)
+
+# 2. Approve: once it validates, a human approves the design.
+wairon lock                 # records the approval in .wai/lock.json
+git add .wai && git commit -m "Approve the design"
+
+# 3. Gate: in CI, check that what merges is the approved design
+#    (or use the reusable workflow — see docs/cli.md#wairon-lock-check---strict).
+wairon lock-check
 
-# Design the system with the SDD architect skill (in your AI tool), or via the
-# sdd_* MCP tools. Then:
+# 4. Implement: your AI tool delegates each component with the sdd-delegate
+#    skill, from its live brief (sdd_get_agent_brief).
 
-wairon status               # spec-tree completeness dashboard
-wairon validate             # architecture-conformance gate
-wairon generate             # regenerate agent files + (re)install skills
-wairon list                 # agents resolved from the spec tree
+# Optional
+wairon export --out design.json   # the whole resolved design as JSON
+wairon diagram                    # interactive canvas in .wai/docs/diagrams/
 ```
 
 ---
@@ -165,20 +203,26 @@ See [docs/cli.md](docs/cli.md). Summary:
 | Command | Description |
 |---------|-------------|
 | `wairon init` | Bootstrap `.wai/` and the SDD spec tree |
-| `wairon status` | Spec-tree completeness dashboard |
-| `wairon validate [--ci]` | Architecture-conformance gate |
-| `wairon generate [--target] [--domain] [--dry-run]` | Regenerate agent files + install skills |
-| `wairon list` / `wairon show <id>` | Inspect agents resolved from the spec tree |
+| `wairon status` | The spec tree, its authoring readiness, and its approval state |
+| `wairon validate [--ci] [--all]` | Architecture-conformance gate (the family run at a project with members) |
+| `wairon doctor [--fix]` | Health check and upgrade repairs |
+| `wairon generate [--target] [--domain] [--dry-run]` | Reconcile guides, skills and context; agent files only when `rules.materializeAgentFiles` is on |
+| `wairon list` / `wairon show <id>` / `wairon agent brief <id>` | Inspect agents resolved from the spec tree; print one's live brief |
+| `wairon export [--out <file>]` | The whole design, resolved, as one JSON document ([format](docs/design-export.md)) |
 | `wairon diagram [--all] [--canvas] [--drawio] [--excalidraw] [--sequence <comp:method>]` | Mermaid, interactive canvas, and editable draw.io/Excalidraw exports |
 | `wairon rules list` | The conformance rule registry (the architecture linter) |
-| `wairon packs add \| list \| remove [--global]` | Extension packs: injected profiles, language tables, and rules |
+| `wairon pack init \| build \| install \| use \| unuse \| impact \| sync \| bundle \| which \| list \| add \| remove` | Extension packs: injected profiles, language tables, and rules |
+| `wairon member …` / `wairon subsystem externalize` / `wairon project rename` | Members (parts and projects) and the family migrations |
+| `wairon externals pin \| status \| list` | Pin and check the externals a project consumes |
 | `wairon domains list \| scan \| add \| remove` | Domains (subsystem-derived + free-standing) |
 | `wairon skills list \| install` | Manage the SDD skills installed into your tools |
-| `wairon lock [-y]` | Validate the design as-complete and record its approval, code findings beside it (generates the implementer agents) |
+| `wairon lock [-y]` | Validate the design as complete and record its approval in `.wai/lock.json`, code findings beside it; no spec file is rewritten |
 | `wairon lock-check [--strict]` | Merge gate: is the design in this tree the design that was approved? Importable as a [reusable workflow](.github/workflows/lock-check.yml) |
 | `wairon mcp serve \| install \| status` | The wairon MCP server (`sdd_*` tools) |
 | `wairon serve [--port] [--data-dir] [--no-auth]` | Self-host: HTTP MCP for many isolated projects + admin plane |
-| `wairon host project \| key \| lock \| promote` | Administer the hosting server (projects, keys, state-scoped lock/promote) |
+| `wairon host unit \| project \| permission \| key \| lock \| doctor \| packs \| git \| producer \| secret \| demo` | Administer the hosting server (units, projects, permissions, keys, the state-scoped lock) |
+| `wairon login` / `wairon remote` / `wairon mcp install --hosted` | Connect a checkout or an AI tool to a hosted instance |
+| `wairon dev` | Local single-project web UI over the current project |
 | `wairon update` / `wairon aliases` | Self-update / command aliases |
 
 ---
@@ -189,7 +233,9 @@ See [docs/cli.md](docs/cli.md). Summary:
 - [Requirements](docs/requirements.md) — goals, non-goals, scope
 - [Roadmap](docs/roadmap.md) — what's done and what's next
 - [Vision](docs/vision.md) — long-term direction
-- [CLI Reference](docs/cli.md) — all commands
+- [CLI Reference](docs/cli.md) — all commands and MCP tools
+- [Design export](docs/design-export.md) — the `wairon export` JSON format for generators and translators
+- [Supervisor doctrine](docs/design/supervisor-doctrine.md) — how Supervisors and Actors may depend on data and effects
 - [Hosted server](docs/design/hosted-mcp-server.md) — self-host wairon over HTTP (Docker, auth, sizing)
 - [Pack scoping](docs/design/pack-scoping.md) — the pack store, per-project selection, and reproducibility (design)
 - [Connecting-agent entrypoint](docs/design/connecting-agent-entrypoint.md) — MCP `instructions`, prompts, and skill composition