@naraya/cli 0.4.0 → 0.4.2

package/README.md

@@ -2,13 +2,12 @@
 
 > One sign-in, every model. A coding agent that delegates specialist work to focused subagents and brings evidence to every claim.
 
-[![Node](https://img.shields.io/badge/node-%E2%89%A522.19-blue)](https://nodejs.org) [![License: MIT](https://img.shields.io/badge/license-MIT-green)](LICENSE) [![Version](https://img.shields.io/badge/version-0.2.4-blue)](package.json)
+[![Node](https://img.shields.io/badge/node-%E2%89%A522.19-blue)](https://nodejs.org) [![License: Proprietary](https://img.shields.io/badge/license-proprietary-red)](LICENSE) [![Version](https://img.shields.io/badge/version-0.4.0-blue)](package.json)
 
-Naraya CLI is a thin wrapper around [pi](https://github.com/earendil-works/pi-mono) that:
+Naraya CLI is an AI coding agent that:
 
-- Routes all model traffic through [router.naraya.ai](https://router.naraya.ai) with one API key.
-- Ships **11 focused subagents** (`nara-build`, `nara-architect`, `nara-release`, `nara-review`, `nara-debug`, `nara-plan`, `nara-explore`, `nara-search`, `nara-fe`, `nara-droid`, `nara-vision`) as folder-based packages with their own skills, scripts, and references.
-- Injects a `APPEND-SYSTEM.md` so the model reaches for the bundled agents and skills on every turn.
+- Routes all model traffic through [router.naraya.ai](https://router.naraya.ai) with one API key — Claude, GPT, Gemini, DeepSeek, GLM, Kimi, MiniMax, Mistral, and more.
+- Ships a roster of focused subagents (`nara-build`, `nara-architect`, `nara-release`, `nara-review`, `nara-debug`, `nara-plan`, `nara-explore`, `nara-search`, `nara-fe`, `nara-droid`) — each with its own skills and references.
 - Provides a `delegate` tool that fans work out to subagents in parallel, sequential chains, or single focus — each in an isolated context window.
 
 ---
@@ -20,10 +19,7 @@ Naraya CLI is a thin wrapper around [pi](https://github.com/earendil-works/pi-mo
 - [Usage](#usage)
 - [Subagents](#subagents)
 - [Skills](#skills)
-- [Agent structure](#agent-structure)
-- [Extending](#extending)
-- [Development](#development)
-- [Architecture](#architecture)
+- [Support](#support)
 - [License](#license)
 
 ---
@@ -35,15 +31,15 @@ Naraya CLI is a thin wrapper around [pi](https://github.com/earendil-works/pi-mo
 npm install -g @naraya/cli
 ```
 
-Or use directly from a checkout:
+Then sign in with your Naraya account:
 
 ```bash
-git clone https://gitlab.naraya.ai/adearman/naracli
-cd naracli
-npm install
-npm link   # expose `naraya` on PATH
+naraya login
 ```
 
+> Naraya CLI is proprietary software. Access requires a valid Naraya account and
+> API key — sign up at [router.naraya.ai](https://router.naraya.ai).
+
 ---
 
 ## Quickstart
@@ -63,7 +59,7 @@ Inside the agent, press `/` for the command palette, `/agents` to list subagents
 
 ```bash
 # Quick subcommands without launching the TUI
-naraya login              # OAuth-style sign-in to router.naraya.ai
+naraya login              # sign in to router.naraya.ai
 naraya logout             # wipe local credentials
 naraya status             # show auth + quota + current model
 naraya status --usd       # same, with USD cost equivalent
@@ -98,17 +94,15 @@ naraya -p "audit the auth module for OWASP top 10"   # single prompt, prints res
 naraya -p --json <task>                              # machine-readable output
 ```
 
-### Delegation patterns (built into the model)
+### Delegation patterns
 
 The agent delegates to specialists via the `delegate` tool. You do not invoke it directly — you phrase the request and the model decides.
 
-Patterns the model uses:
-
 ```ts
 // One focused specialist
 delegate { agent: "nara-droid", task: "..." }
 
-// Independent units in parallel (max 8 at once, 4 concurrent)
+// Independent units in parallel
 delegate { tasks: [
   { agent: "nara-search",  task: "research X" },
   { agent: "nara-explore", task: "find usages of Y" },
@@ -127,268 +121,64 @@ delegate { chain: [
 
 ## Subagents
 
-Naraya ships **11 subagents**. Each is a self-contained folder under `assets/agents/`.
+Naraya ships a roster of focused subagents. The model picks based on the **unit of work**, not the surface form.
 
 ### Coordination
 
 | Agent | Purpose | When to use |
 |---|---|---|
 | `nara-build` | Implements features, bugfixes, refactors, releases | Top-level orchestrator. Default for "implement X", "fix Y", "refactor Z" |
-| `nara-architect` | System design, ADRs, component decomposition | Greenfield design, scaling decisions, API contract design, trade-off analysis. Read-only — never edits code |
-| `nara-release` | Version sync, verification, deploy safety | Before any production push, mobile store release, or schema migration |
-| `nara-review` | Multi-aspect code review (security, perf, a11y, API, tests) | Pre-merge review, audit, self-review before push. Does not fix |
-| `nara-debug` | 4-phase root-cause debugging | "X is broken", flaky tests, intermittent failures. Stops after 3 failed fix attempts |
+| `nara-architect` | System design, ADRs, component decomposition | Greenfield design, scaling decisions, API contracts, trade-offs. Read-only |
+| `nara-release` | Version sync, verification, deploy safety | Before any production push, store release, or schema migration |
+| `nara-review` | Multi-aspect code review (security, perf, a11y, API, tests) | Pre-merge review, audit, self-review. Does not fix |
+| `nara-debug` | 4-phase root-cause debugging | "X is broken", flaky tests, intermittent failures |
 
 ### Investigation (read-only)
 
 | Agent | Purpose | When to use |
 |---|---|---|
 | `nara-plan` | Implementation plans, never edits code | Pair with `nara-build` for hand-off |
-| `nara-explore` | Fast codebase navigation (haiku) | "Where is X defined?", "who calls Y?", quick map |
-| `nara-search` | Evidence-first research with citations | Library comparison, version migration, external doc lookup |
+| `nara-explore` | Fast codebase navigation | "Where is X defined?", "who calls Y?", quick map |
+| `nara-search` | Evidence-first research with citations | Library comparison, version migration, doc lookup |
 
 ### Specialist (writes code in their domain)
 
 | Agent | Purpose | When to use |
 |---|---|---|
 | `nara-fe` | Frontend UI/UX (React, Vue, Svelte, Angular, CSS, Tailwind, a11y) | UI changes, component design, responsive, accessibility |
-| `nara-droid` | Native Android (Kotlin, Gradle, Compose, Room, Hilt, adb, APK/AAB) | Android-specific tasks, build failures, release |
-| `nara-vision` | Image analysis using mistral-medium-3-5 vision model | Screenshots, diagrams, UI layouts, code snippets, visual content |
-
-### Picking the right agent
+| `nara-droid` | Native Android (Kotlin, Gradle, Compose, Room, Hilt, adb, APK/AAB) | Android tasks, build failures, release |
 
-The model picks based on the **unit of work**, not the surface form. Quick heuristic:
-
-- The user said "implement", "fix", "build", "add", "refactor" → `nara-build`
-- The user said "plan", "design", "architect", "decompose" → `nara-architect` (design) or `nara-plan` (task breakdown)
-- The user said "review", "audit", "check this" → `nara-review`
-- The user said "release", "deploy", "ship" → `nara-release`
-- The user said "debug", "broken", "crash", "flaky" → `nara-debug`
-- The user said "find", "where", "who calls" → `nara-explore`
-- The user said "research", "compare", "what is the best", "how does X work in library Y" → `nara-search`
-- The user said something Android-specific → `nara-droid`
-- The user said something UI/frontend-specific → `nara-fe`
-- The user shared an image, screenshot, or needs visual analysis → `nara-vision`
-
-If multiple of these apply, the model fans out to parallel subagents.
+If multiple roles apply, the model fans out to parallel subagents.
 
 ---
 
 ## Skills
 
-Naraya bundles a **global skill pool** at `skills/` (~100 skills) that any agent can load on demand via `/skill:<name>` or by reading `skills/<name>/SKILL.md` directly.
-
-Highlights:
+Naraya bundles a large pool of skills that any agent can load on demand via `/skill:<name>`. Highlights:
 
 | Skill | What it covers |
 |---|---|
 | `systematic-debugging` | 4-phase root-cause debugging |
 | `test-driven-development` | RED-GREEN-REFACTOR discipline |
 | `writing-plans` | Implementation plans with bite-sized tasks |
-| `web-app-release-workflow` | Ship Dockerized web apps |
 | `security`, `auth-identity`, `compliance-governance` | Security review and compliance |
 | `architecture`, `api-design-patterns`, `distributed-systems` | System design |
-| `mlops/*` | ML/LLM training, serving, fine-tuning, evaluation |
-| `creative/{humanizer, p5js, claude-design}` | Content / UI generation |
 | `android-kotlin`, `compose-patterns`, `gradle-troubleshoot` | Android development |
 | `react-patterns`, `css-patterns` | Frontend |
 
 **Skills vs subagents:** a skill is a workflow / lens the agent applies inline. A subagent is a separate context that runs in parallel. Use skills for "do X the right way", subagents for "do X in isolation".
 
----
-
-## Agent structure
-
-Each agent is a folder under `assets/agents/`. Folder name MUST equal the frontmatter `name:` field.
-
-```
-assets/agents/
-├── nara-build/                 # Generalist implementer (top-level)
-│   ├── AGENTS.md               # Frontmatter (name/description/tools/model) + system prompt
-│   ├── skills/                 # Agent-specific skills, NOT published to global pool
-│   │   └── engineering-patterns/SKILL.md
-│   ├── scripts/                # Helper shell scripts (chmod +x)
-│   │   ├── impact-scan.sh
-│   │   └── scan.sh
-│   ├── references/             # On-demand docs the agent may read into context
-│   │   └── adr-template.md
-│   ├── assets/                 # Static resources (templates, data)
-│   └── README.md               # per-agent docs
-├── nara-plan/                  # Read-only planner
-├── nara-explore/               # Fast read-only code navigation (haiku)
-├── nara-search/                # Evidence-first research
-├── nara-droid/                 # Native Android specialist
-├── nara-fe/                    # Frontend UI/UX specialist
-├── nara-vision/                # Image analysis (mistral-medium-3-5 vision)
-├── nara-architect/             # System design & ADRs (read-only)
-├── nara-release/               # Release safety
-├── nara-review/                # Multi-aspect code review
-└── nara-debug/                 # 4-phase root-cause debugging
-```
-
-### Conventions
-
-- **Folder name = frontmatter `name:`.** The orchestrator warns but does not fail if they diverge.
-- **`AGENTS.md` is the only file the loader reads.** Everything else is opt-in for the agent to `read` into context.
-- **Per-agent `skills/` are NOT published to the global skill pool.** They are scoped to that agent.
-- **Scripts are executable** (`chmod +x`). They can be invoked from the agent's `bash` tool.
-- **References are docs the agent may read.** Large checklists, protocol walkthroughs, template snippets.
-
-### Adding a new agent
-
-1. Create `assets/agents/<your-role>/AGENTS.md` with frontmatter (`name`, `description`, `tools`, `model`) + body.
-2. Add the four subdirs: `skills/`, `scripts/`, `references/`, `assets/`.
-3. (Optional) Add a `README.md` documenting the agent.
-4. Run `npm test` to confirm the seed manifest picks it up.
-5. (Optional) Add the agent to the roster in `assets/APPEND-SYSTEM.md` so the model knows to delegate to it.
-
-### Editing an existing agent
-
-Edit `assets/agents/<role>/AGENTS.md`. The orchestrator hot-reads this on each `delegate` call, so changes take effect on the next invocation. No restart needed.
-
----
-
-## Extending
-
-### Adding a skill to a specific agent
-
-Drop a directory under `assets/agents/<role>/skills/<skill-name>/` with a `SKILL.md` inside:
-
-```
-assets/agents/nara-fe/
-└── skills/
-    └── vue-patterns/
-        └── SKILL.md
-```
-
-### Adding a skill to the global pool
-
-Drop it under `skills/<skill-name>/SKILL.md` at the repo root. The global pool is auto-discovered by pi.
-
-### Overriding model per-agent
-
-Edit the frontmatter `model:` field in `AGENTS.md`. Or set a per-agent override at runtime with `/agent-model <agent> <model>`.
-
-### Customizing the system prompt
-
-Edit `assets/APPEND-SYSTEM.md`. This file is appended to the model's system prompt on every launch.
-
----
-
-## Development
-
-```bash
-git clone https://gitlab.naraya.ai/adearman/naracli
-cd naracli
-npm install
-npm test                # run node:test suite (8 tests)
-```
-
-### Project layout
-
-```
-naraya-cli/
-├── bin/
-│   └── naraya.mjs             # CLI entry: subcommand dispatch + pi launch
-├── src/
-│   ├── banner.mjs             # login/launch banner
-│   ├── clipboard.mjs          # copy auth URL to clipboard
-│   ├── config.mjs             # models.json read/write
-│   ├── goodbye.mjs            # farewell on quit
-│   ├── login.mjs              # naraya login (OAuth-style)
-│   ├── seed.mjs               # asset sync to ~/.naraya/agent
-│   ├── splash.mjs             # startup animation
-│   └── status.mjs             # naraya status (quota, cost)
-├── assets/
-│   ├── APPEND-SYSTEM.md       # injected into system prompt
-│   ├── agents/                # 10 subagent folders (see above)
-│   ├── extensions/            # .ts extensions loaded by pi
-│   ├── skills/                # empty (skills live at repo root for global pool)
-│   └── themes/naraya.json     # default theme
-├── skills/                    # global skill pool (~100 skills)
-├── test/
-│   ├── config.test.mjs
-│   ├── seed.test.mjs          # covers agent folder + nested resources
-│   └── status.test.mjs
-├── package.json
-└── README.md
-```
-
-### Testing
-
-```bash
-npm test                                  # all 8 tests
-node --test test/seed.test.mjs           # just seed (asset sync)
-node --test test/status.test.mjs         # just status rendering
-```
-
-The seed test covers both the legacy flat layout and the new folder-based agent layout (AGENTS.md + nested skills/scripts).
+For project-specific guidance, add an `AGENTS.md` to your repository — the agent reads it automatically.
 
 ---
 
-## Architecture
-
-### Request flow
-
-```
-User input
-   │
-   ▼
-┌─────────────────────────┐
-│ bin/naraya.mjs          │  CLI entry, subcommand dispatch
-│  - login/logout/status  │
-│  - else: launch pi      │
-└──────────┬──────────────┘
-           │
-           ▼
-┌─────────────────────────┐
-│ src/seed.mjs            │  Sync assets/ to ~/.naraya/agent
-│  - assets/agents/*      │   (only managed files; user files preserved)
-│  - assets/extensions/*  │
-│  - assets/APPEND-SYSTEM │
-└──────────┬──────────────┘
-           │
-           ▼
-┌─────────────────────────┐
-│ pi-coding-agent         │  pi runtime
-│  - system prompt:       │   ← APPEND-SYSTEM.md appended
-│  - skill discovery      │   ← global pool + per-agent
-│  - extension loading    │   ← naraya-orchestrator, naraya-debug, etc.
-│  - delegate tool        │   ← fans out to subagents
-└──────────┬──────────────┘
-           │
-           ▼
-┌─────────────────────────┐
-│ Subagent (isolated ctx) │  one per `delegate` call
-│  - reads AGENTS.md      │
-│  - reads per-agent      │
-│    skills/scripts/refs  │
-│  - reads global skills  │
-│  - returns result       │
-└─────────────────────────┘
-           │
-           ▼
-        Output to user
-```
-
-### Why folder-based agents?
-
-- **Colocation.** Each agent is one self-contained directory: definition + skills + scripts + references + assets. Move it, version it, delete it as one unit.
-- **No global namespace pollution.** Per-agent `skills/` are scoped; only `AGENTS.md` is published.
-- **Easier to share.** A folder is a package. An `.md` file in a flat dir is a file in a list.
-- **Scales.** Adding an agent = adding a folder. Adding an agent to the old flat layout = adding a file that could collide.
-- **Backward compatible.** The orchestrator loader supports both folder layout (`<role>/AGENTS.md`) and legacy flat (`<role>.md`).
-
-### Why split coordination / specialist / investigation?
-
-- **Coordination agents** (`nara-build`, `nara-architect`, `nara-release`, `nara-review`, `nara-debug`) need broad context. They route, sequence, verify.
-- **Specialist agents** (`nara-fe`, `nara-droid`) need deep domain knowledge. Their system prompt is shaped by their domain.
-- **Investigation agents** (`nara-plan`, `nara-explore`, `nara-search`) are read-only. They never modify state. They produce artifacts (plans, maps, evidence) that the orchestrator consumes.
+## Support
 
-Mixing these roles in one agent produces shallow specialists and shallow coordinators.
+- **Account, billing, API keys:** [router.naraya.ai](https://router.naraya.ai)
+- **Sign-in issues:** run `naraya logout` then `naraya login` to refresh your session.
 
 ---
 
 ## License
 
-MIT. See [LICENSE](LICENSE).
+Proprietary. Copyright © 2026 PT. Naraya Teknologi Indonesia. All rights reserved. See [LICENSE](LICENSE).

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@naraya/cli",
-  "version": "0.4.0",
+  "version": "0.4.2",
   "description": "Naraya AI coding agent - one sign-in, all models via router.naraya.ai",
   "type": "module",
   "bin": {