profclaw 0.0.1-beta.2 → 2.0.0

package/CHANGELOG.md

@@ -1,49 +1,32 @@
 # Changelog
 
-## [2.0.0](https://github.com/profclaw/profclaw/compare/v1.0.0...v2.0.0) (2026-03-11)
+All notable changes to this project will be documented in this file.
 
-### Breaking Changes
+## [2.0.0](https://github.com/profclaw/profclaw/releases/tag/v2.0.0) (2026-03-14)
 
-* Renamed from "GLINR Task Manager" to "profClaw" — all env vars, CLI commands, and API paths updated
-* `GLINR_REQUIRE_REDIS` renamed to `PROFCLAW_REQUIRE_REDIS`
-* `glinr.mjs` renamed to `profclaw.mjs`
-* All `glinr-*` skill names renamed to `profclaw-*`
+Initial public beta release.
 
 ### Features
 
-* **Deployment Modes**: Run as `pico`, `mini`, or `pro` via `PROFCLAW_MODE` env var
-* **In-Memory Queue**: BullMQ-compatible task queue without Redis for pico/mini modes
-* **Unified Queue Facade**: Auto-selects Redis or memory backend based on mode
-* **WebChat Provider**: Browser-based chat via SSE with session management
-* **Matrix Provider**: Matrix homeserver integration via Client-Server API
-* **Google Chat Provider**: Google Workspace integration via webhook or service account
-* **MS Teams Provider**: Microsoft Teams integration via Bot Framework
-* **Plugin SDK**: Third-party plugin system with `definePlugin()` helper, npm/local loader
-* **Onboarding CLI**: `npx profclaw onboard` wizard for zero-to-running setup
-* **Skills Expansion**: 15 built-in skills (code-review, code-generation, debug-helper, git-workflow, web-research, cron-manager, memory-manager, system-admin, file-manager, api-tester, docker-ops)
-* **Sandbox Hardening**: Security presets per deployment mode
-* **Static UI Serving**: Hono middleware serves `ui/dist/` with SPA fallback
-* **Pico Docker Image**: Minimal image without Chromium (`Dockerfile.pico`)
+* **35 AI providers** - Anthropic, OpenAI, Google, Groq, Ollama, DeepSeek, Bedrock, and more
+* **22 chat channels** - Slack, Discord, Telegram, WhatsApp, iMessage, Matrix, Teams, and 15 more
+* **72 built-in tools** - file ops, git, browser automation, cron, web search, canvas, voice
+* **50 skills** - coding agent, GitHub issues, Notion, Obsidian, image gen, and more
+* **3 deployment modes** - pico (~140MB), mini (~145MB), pro (full features)
+* **MCP server** - native Model Context Protocol support (stdio + SSE)
+* **Voice I/O** - STT (Whisper) + TTS (ElevenLabs/OpenAI/system) + Talk Mode
+* **Plugin SDK** - build and share third-party plugins via npm
+* **Mode-aware routing** - API endpoints, tools, and capabilities gated by deployment mode
+* **In-memory queue** - BullMQ-compatible task queue without Redis for pico/mini modes
+* **Cost tracking** - per-token budget management with alerts at 50/80/100%
+* **Project management** - tickets, Kanban, sprints, burndown charts, GitHub/Jira/Linear sync
 
-### Chat Channels (4 → 8)
+### Chat Channels
 
-* Slack, Discord, Telegram, WhatsApp (existing)
-* WebChat, Matrix, Google Chat, Microsoft Teams (new)
+Slack, Discord, Telegram, WhatsApp, WebChat, Matrix, Google Chat, Microsoft Teams, iMessage, Signal, IRC, LINE, Mattermost, DingTalk, WeCom, Feishu/Lark, QQ, Nostr, Twitch, Zalo, Nextcloud Talk, Synology Chat
 
-### Infrastructure
+### Integrations
 
-* CI/CD: GitHub Actions with build+test+lint, release-please, Docker multi-arch, npm publish
-* OSS files: CONTRIBUTING.md, SECURITY.md, CODE_OF_CONDUCT.md, LICENSE (AGPL-3.0)
-
-## [1.0.0](https://github.com/profclaw/profclaw/releases/tag/v1.0.0) (2026-03-10)
-
-### Features
-
-* Initial release as GLINR Task Manager v1.0.0
-* AI agent task orchestration with 51+ built-in tools
-* 15+ AI provider support (Anthropic, OpenAI, Google, Azure, Ollama, etc.)
-* 4 chat channels (Slack, Discord, Telegram, WhatsApp)
-* GitHub, Jira, Linear integrations
-* MCP server for Model Context Protocol
-* Redis-backed BullMQ task queue
-* React 19 dashboard UI
+* GitHub (webhooks, OAuth, issue sync, PR automation)
+* Jira (webhooks, OAuth, issue sync, transitions)
+* Linear (webhooks, OAuth, issue sync)

package/CONTRIBUTING.md

@@ -50,7 +50,7 @@ Open a discussion or issue describing:
 
 1. Fork the repo and create a branch from `main`
 2. Follow the naming convention: `feat/description`, `fix/description`, `refactor/description`
-3. Write code following existing patterns (see `CLAUDE.md`)
+3. Write code following existing patterns (see `AGENTS.md` for conventions)
 4. Add tests for new features
 5. Ensure `pnpm build && pnpm test && pnpm lint` pass
 6. Submit a pull request
@@ -61,8 +61,6 @@ We use conventional commits:
 
 ```
 <type>: <description>
-
-Co-Authored-By: Your Name <your@email.com>
 ```
 
 Types: `feat`, `fix`, `perf`, `refactor`, `docs`, `test`
@@ -75,7 +73,7 @@ Types: `feat`, `fix`, `perf`, `refactor`, `docs`, `test`
 - Tailwind v4 for UI styling (no inline styles)
 - Lucide icons only
 - Zod validation at system boundaries
-- See `CLAUDE.md` for full conventions
+- See `AGENTS.md` for full conventions
 
 ## Architecture
 
@@ -86,7 +84,7 @@ src/
   core/          - Deployment modes, feature flags
   integrations/  - GitHub, Jira, Linear webhooks
   queue/         - BullMQ + in-memory task queue
-  providers/     - AI SDK providers (15+)
+  providers/     - AI SDK providers (35+)
   skills/        - Skill loader and registry
   types/         - Shared TypeScript types
 

package/README.md

@@ -1,6 +1,12 @@
 <p align="center">
-  <h1 align="center">profClaw</h1>
-  <p align="center">Smart, lightweight AI agent engine. Run locally on anything from a $10 board to a VPS.</p>
+  <img src="ui/public/brand/profclaw-appicon.svg" alt="profClaw" width="120" height="120">
+</p>
+
+<h1 align="center">profClaw</h1>
+
+<p align="center">
+  <b>Smart, lightweight AI agent engine.</b><br>
+  Run locally on anything from a Raspberry Pi to a production VPS.
 </p>
 
 <p align="center">
@@ -18,9 +24,37 @@
 
 ---
 
-profClaw is a local-first AI agent that talks to your tools. Connect it to Slack, Discord, Telegram, or any of 22 supported chat channels. Point it at your GitHub, Jira, or Linear. Give it an AI provider (Anthropic, OpenAI, Ollama, or 30+ others). It handles the rest - executing tasks, managing workflows, and automating your dev life through natural language.
+## Table of Contents
+
+- [Why profClaw](#why-profclaw)
+- [Features](#features)
+- [Installation](#installation)
+- [Deployment Modes](#deployment-modes)
+- [Where It Runs](#where-it-runs)
+- [Configuration](#configuration)
+- [AI Providers](#ai-providers)
+- [Chat Channels](#chat-channels)
+- [Integrations](#integrations)
+- [Architecture](#architecture)
+- [Development](#development)
+- [License](#license)
+
+---
+
+## Why profClaw
+
+Most AI coding tools are either cloud-only (Cursor, Devin) or single-purpose CLIs (Aider, SWE-Agent). profClaw is different:
 
-Runs on your machine. Your data stays local. No cloud required.
+- **Local-first** - runs on your machine, your data stays local
+- **Multi-agent orchestration** - routes tasks to the right agent (Claude, GPT, Ollama) based on capability scoring, not just round-robin
+- **Real task queue** - BullMQ with dead letter queue, retry with backoff, priority scheduling
+- **Built-in project management** - tickets, Kanban, sprints, burndown charts, bi-directional sync with GitHub/Jira/Linear
+- **22 chat channels** - talk to your AI through Slack, Discord, Telegram, WhatsApp, Teams, or 17 others
+- **Cost tracking** - per-token budget management with alerts at 50/80/100%
+- **72 built-in tools** - file ops, git, browser automation, cron, web search, canvas, voice
+- **Mode-aware** - scales from a Raspberry Pi (pico) to a full team server (pro)
+
+No other open-source tool combines task orchestration, project management, and cost tracking in one self-hosted package.
 
 ## Features
 
@@ -33,30 +67,24 @@ Runs on your machine. Your data stays local. No cloud required.
 | **MCP server** | Native Model Context Protocol - connect to any MCP-compatible client |
 | **Voice I/O** | STT (Whisper) + TTS (ElevenLabs/OpenAI/system) + Talk Mode |
 | **Plugin SDK** | Build and share third-party plugins via npm or ClawHub |
-| **3 deployment modes** | Pico (~140MB), Mini (~180MB), Pro (full features) |
+| **3 deployment modes** | Pico (~140MB), Mini (~145MB), Pro (full features) |
 
-## Deployment Modes
+## Installation
 
-profClaw scales from a Raspberry Pi to a full production server:
+### npm (recommended)
 
-| Mode | What you get | RAM | Best for |
-|------|-------------|-----|----------|
-| **pico** | Agent + tools + chat. No UI. | ~140MB | IoT, home automation, $10 boards |
-| **mini** | + Dashboard, basic cron, 1-2 channels | ~180MB | Personal dev server, VPS |
-| **pro** | + All channels, Redis queues, plugins, sandbox | ~200MB | Teams, production |
-
-Set via `PROFCLAW_MODE=pico|mini|pro` environment variable.
+```bash
+npx profclaw onboard
+```
 
-## Quick Start
+This runs the zero-to-running wizard: picks your AI provider, sets up config, and starts the server.
 
-### npm
+Or install globally:
 
 ```bash
-npx profclaw onboard  # Zero-to-running wizard
-# or
 npm install -g profclaw@latest
-profclaw setup        # Interactive config
-profclaw serve        # Start the agent
+profclaw setup
+profclaw serve
 ```
 
 ### Docker
@@ -74,11 +102,11 @@ docker run -d \
 ```bash
 git clone https://github.com/profclaw/profclaw.git
 cd profclaw
-cp .env.example .env    # Add your API key
+cp .env.example .env
 docker compose up -d
 ```
 
-**Free local AI?** Add Ollama:
+Want free local AI? Add Ollama:
 
 ```bash
 docker compose --profile ai up -d
@@ -90,6 +118,32 @@ docker compose --profile ai up -d
 curl -fsSL https://raw.githubusercontent.com/profclaw/profclaw/main/install.sh | bash
 ```
 
+## Deployment Modes
+
+profClaw scales from a Raspberry Pi to a full production server:
+
+| Mode | What you get | RAM | Best for |
+|------|-------------|-----|----------|
+| **pico** | Agent + tools + 1 chat channel + cron. No UI. | ~140MB | Raspberry Pi, $5 VPS, home server |
+| **mini** | + Dashboard, integrations, 3 channels | ~145MB | Personal dev server, small VPS |
+| **pro** | + All channels, Redis queues, plugins, browser tools | ~120-200MB | Teams, production |
+
+Set via `PROFCLAW_MODE=pico|mini|pro` environment variable.
+
+## Where It Runs
+
+| Hardware | RAM | Recommended mode |
+|----------|-----|-----------------|
+| Raspberry Pi Zero 2W | 512MB | pico |
+| Raspberry Pi 3/4/5 | 1-8GB | mini or pro |
+| Orange Pi / Rock Pi | 1-4GB | mini or pro |
+| $5/mo VPS (Hetzner, OVH) | 512MB-1GB | pico or mini |
+| Old laptop / home PC | 4-16GB | pro |
+| Docker (alongside other services) | 512MB+ | any mode |
+| Old Android phone (Termux) | 1-2GB | pico |
+
+profClaw requires Node.js 22+. For bare-metal embedded devices (ESP32, Arduino), see [MimiClaw](https://github.com/mimiclaw) (C) or [PicoClaw](https://github.com/picoclaw) (Go).
+
 ## Configuration
 
 The setup wizard (`profclaw setup`) handles everything interactively. Or set environment variables:
@@ -101,8 +155,8 @@ OPENAI_API_KEY=sk-xxx
 OLLAMA_BASE_URL=http://localhost:11434
 
 # Deployment
-PROFCLAW_MODE=mini          # pico | mini | pro
-PORT=3000                   # HTTP server port
+PROFCLAW_MODE=mini
+PORT=3000
 
 # Optional
 REDIS_URL=redis://localhost:6379   # Required for pro mode
@@ -110,26 +164,6 @@ REDIS_URL=redis://localhost:6379   # Required for pro mode
 
 See [`.env.example`](.env.example) for all options.
 
-## Architecture
-
-```
-src/
-  adapters/       AI agent adapters (tool calling, streaming)
-  chat/           Chat engine + execution pipeline
-    providers/    Slack, Discord, Telegram, WhatsApp, WebChat, ...
-    execution/    Tool executor, sandbox, agentic loop
-  core/           Deployment modes, feature flags
-  integrations/   GitHub, Jira, Linear webhooks
-  queue/          BullMQ (pro) + in-memory (pico/mini) task queue
-  providers/      35 AI SDK providers
-  skills/         Skill loader and registry
-  mcp/            MCP server (stdio + SSE)
-  types/          Shared TypeScript types
-
-ui/src/           React 19 + Vite dashboard (mini/pro only)
-skills/           Built-in skill definitions
-```
-
 ## AI Providers
 
 35 providers with 90+ model aliases:
@@ -145,15 +179,38 @@ skills/           Built-in skill definitions
 | DeepSeek | V3, R1 | No |
 | Azure OpenAI | GPT-4o | No |
 | xAI | Grok | No |
-| Zhipu AI | GLM-4, GLM-5 | No |
-| Moonshot | Kimi | No |
-| Qwen API | Qwen 2.5 | No |
+| OpenRouter | Any model | No |
 | Together | Open models | No |
 | Fireworks | Open models | No |
 | Mistral | Mistral Large, Codestral | No |
-| OpenRouter | Any model | No |
-| Replicate | Open models | No |
-| ... and 18 more | HuggingFace, NVIDIA NIM, Watsonx, Volcengine, etc. | |
+| ... and 22 more | HuggingFace, NVIDIA NIM, Cerebras, Replicate, Zhipu, Moonshot, Qwen, etc. | |
+
+## Chat Channels
+
+| Channel | Protocol |
+|---------|----------|
+| Slack | Bolt SDK |
+| Discord | HTTP Interactions |
+| Telegram | Bot API |
+| WhatsApp | Cloud API |
+| WebChat | SSE (browser-based) |
+| Matrix | Client-Server API |
+| Google Chat | Webhook + API |
+| Microsoft Teams | Bot Framework |
+| iMessage | BlueBubbles |
+| Signal | signald bridge |
+| IRC | TLS, RFC 1459 |
+| LINE | Messaging API |
+| Mattermost | REST API v4 |
+| DingTalk | OpenAPI + webhook |
+| WeCom | WeChat Work API |
+| Feishu/Lark | Open Platform |
+| QQ | Bot API |
+| Nostr | Relay protocol |
+| Twitch | Helix API + IRC |
+| Zalo | OA API v3 |
+| Nextcloud Talk | OCS API |
+| Synology Chat | Webhook |
 
 ## Integrations
 
@@ -162,14 +219,26 @@ skills/           Built-in skill definitions
 | **GitHub** | Webhooks, OAuth, issue sync, PR automation |
 | **Jira** | Webhooks, OAuth, issue sync, transitions |
 | **Linear** | Webhooks, OAuth, issue sync |
-| **Slack** | Slash commands, DM pairing, thread replies |
-| **Discord** | Bot commands, channel routing |
-| **Telegram** | Bot API, inline commands |
-| **WhatsApp** | Cloud API or Baileys |
-| **WebChat** | Browser-based SSE chat, embeddable widget |
-| **Matrix** | Room-based, Client-Server API |
-| **Google Chat** | Webhook + service account, Cards v2 |
-| **MS Teams** | Bot Framework, Adaptive Cards |
+
+## Architecture
+
+```
+src/
+  adapters/       AI agent adapters (tool calling, streaming)
+  chat/           Chat engine + execution pipeline
+    providers/    Slack, Discord, Telegram, WhatsApp, WebChat, ...
+    execution/    Tool executor, sandbox, agentic loop
+  core/           Deployment modes, feature flags
+  integrations/   GitHub, Jira, Linear webhooks
+  queue/          BullMQ (pro) + in-memory (pico/mini) task queue
+  providers/      35 AI SDK providers
+  skills/         Skill loader and registry
+  mcp/            MCP server (stdio + SSE)
+  types/          Shared TypeScript types
+
+ui/src/           React 19 + Vite dashboard (mini/pro only)
+skills/           Built-in skill definitions
+```
 
 ## Development
 
@@ -187,10 +256,6 @@ See [CONTRIBUTING.md](CONTRIBUTING.md) for full guidelines.
 
 See [SECURITY.md](SECURITY.md) for our security policy and how to report vulnerabilities.
 
-## Docs & Website
-
-Documentation and marketing site live in a separate repo: [`profclaw-site`](https://github.com/profclaw/profclaw-site)
-
 ## License
 
-[AGPL-3.0](LICENSE) -- profclaw contributors, 2024-2026
+[AGPL-3.0](LICENSE)

package/dist/chat/execution/tools/index.d.ts

@@ -58,7 +58,7 @@ export type { SessionsSendParams, SessionsSendResult } from './sessions-send.js'
 export type { BrowserNavigateResult, BrowserSnapshotResult, BrowserClickResult, BrowserTypeResult, BrowserSearchResult, BrowserScreenshotResult, BrowserPagesResult, BrowserCloseResult, } from './browser.js';
 export declare const builtinTools: ToolDefinition[];
 /**
- * Register all built-in tools (with tier assignments)
+ * Register all built-in tools (with tier assignments and mode gating)
  */
 export declare function registerBuiltinTools(): void;
 //# sourceMappingURL=index.d.ts.map

package/dist/chat/execution/tools/index.js

@@ -4,6 +4,7 @@
  * Export all built-in tools and registration function.
  */
 import { getToolRegistry } from '../registry.js';
+import { hasCapability } from '../../../core/deployment.js';
 // Core tools
 import { execTool } from './exec.js';
 import { webFetchTool } from './web-fetch.js';
@@ -233,12 +234,50 @@ const rawBuiltinTools = [
 ];
 // Apply tier assignments to all tools
 export const builtinTools = applyTiers(rawBuiltinTools);
+/** Tool names that require specific deployment mode capabilities */
+const BROWSER_TOOL_NAMES = new Set([
+    'browser_navigate', 'browser_snapshot', 'browser_click', 'browser_type',
+    'browser_search', 'browser_screenshot', 'browser_pages', 'browser_close',
+]);
+const CRON_TOOL_NAMES = new Set([
+    'cron_create', 'cron_list', 'cron_trigger', 'cron_pause',
+    'cron_archive', 'cron_delete', 'cron_history',
+]);
+const CHANNEL_TOOL_NAMES = new Set([
+    'discord_actions', 'slack_actions', 'telegram_actions',
+]);
+const INTEGRATION_TOOL_NAMES = new Set([
+    'create_ticket', 'list_tickets', 'update_ticket', 'get_ticket',
+    'create_project', 'list_projects',
+]);
 /**
- * Register all built-in tools (with tier assignments)
+ * Register all built-in tools (with tier assignments and mode gating)
  */
 export function registerBuiltinTools() {
     const registry = getToolRegistry();
-    for (const tool of builtinTools) {
+    const modeFilteredTools = builtinTools.filter((tool) => {
+        // Browser tools: pro only (browser_tools capability)
+        if (BROWSER_TOOL_NAMES.has(tool.name))
+            return hasCapability('browser_tools');
+        // Cron tools: mini+ (cron capability)
+        if (CRON_TOOL_NAMES.has(tool.name))
+            return hasCapability('cron');
+        // Channel action tools: mini+ (chat_channels capability)
+        if (CHANNEL_TOOL_NAMES.has(tool.name))
+            return hasCapability('chat_channels');
+        // Integration tools: mini+ (integrations capability)
+        if (INTEGRATION_TOOL_NAMES.has(tool.name))
+            return hasCapability('integrations');
+        // Screen capture: pro only (sandbox_full capability)
+        if (tool.name === 'screen_capture')
+            return hasCapability('sandbox_full');
+        // Canvas render: pro only (browser_tools - needs headless browser)
+        if (tool.name === 'canvas_render')
+            return hasCapability('browser_tools');
+        // Everything else: always available
+        return true;
+    });
+    for (const tool of modeFilteredTools) {
         registry.register(tool);
     }
 }

package/dist/chat/providers/slack/index.d.ts

@@ -29,9 +29,9 @@ declare const SlackAccountConfigSchema: z.ZodObject<{
     teamId?: string | undefined;
     teamName?: string | undefined;
     botToken?: string | undefined;
-    webhookUrl?: string | undefined;
     appToken?: string | undefined;
     signingSecret?: string | undefined;
+    webhookUrl?: string | undefined;
 }, {
     id: string;
     provider: "slack";
@@ -42,9 +42,9 @@ declare const SlackAccountConfigSchema: z.ZodObject<{
     teamId?: string | undefined;
     teamName?: string | undefined;
     botToken?: string | undefined;
-    webhookUrl?: string | undefined;
     appToken?: string | undefined;
     signingSecret?: string | undefined;
+    webhookUrl?: string | undefined;
 }>;
 export declare const slackProvider: ChatProvider<SlackAccountConfig>;
 export { SlackAccountConfigSchema };

package/dist/cli/commands/models.js

@@ -16,6 +16,13 @@ function formatCostPair(input, output) {
         return chalk.dim('-');
     return `$${input.toFixed(2)}/$${output.toFixed(2)}`;
 }
+/**
+ * Extract the provider type from a model ID (e.g. "anthropic/claude-3" -> "anthropic")
+ */
+function extractProvider(modelId) {
+    const slash = modelId.indexOf('/');
+    return slash > 0 ? modelId.slice(0, slash) : undefined;
+}
 export function modelsCommands() {
     const cmd = new Command('models')
         .description('Manage AI model configuration');
@@ -29,13 +36,13 @@ export function modelsCommands() {
         const spin = spinner('Fetching models...').start();
         try {
             const params = options.provider ? `?provider=${encodeURIComponent(options.provider)}` : '';
-            const result = await api.get(`/api/providers/models${params}`);
+            const result = await api.get(`/api/chat/models${params}`);
             spin.stop();
             if (!result.ok) {
                 error(result.error || 'Failed to fetch models');
                 process.exit(1);
             }
-            const { models, default: defaultModel } = result.data;
+            const { models } = result.data;
             if (options.json) {
                 console.log(JSON.stringify(result.data, null, 2));
                 return;
@@ -44,21 +51,17 @@ export function modelsCommands() {
                 info('No models available.');
                 return;
             }
-            const table = createTable(['Model', 'Provider', 'Context', 'Cost (in/out)', 'Default']);
+            const table = createTable(['Model', 'Provider', 'Context', 'Cost (in/out)']);
             for (const m of models) {
-                const isDefault = m.id === defaultModel || m.isDefault;
                 table.push([
-                    isDefault ? chalk.cyan(truncate(m.id, 35)) + chalk.yellow(' *') : truncate(m.id, 35),
+                    truncate(m.id, 40),
                     m.provider,
                     formatContext(m.contextWindow),
                     formatCostPair(m.costPer1MInput, m.costPer1MOutput),
-                    isDefault ? chalk.yellow('yes') : chalk.dim('no'),
                 ]);
             }
             console.log(table.toString());
             console.log(chalk.dim(`\n${models.length} model${models.length !== 1 ? 's' : ''}`));
-            if (defaultModel)
-                console.log(chalk.dim(`Default: ${defaultModel}`));
         }
         catch (err) {
             spin.stop();
@@ -73,13 +76,17 @@ export function modelsCommands() {
         .action(async (model, options) => {
         const spin = spinner(`Fetching info for ${model}...`).start();
         try {
-            const result = await api.get(`/api/providers/models/${encodeURIComponent(model)}`);
+            const result = await api.get('/api/chat/models');
             spin.stop();
             if (!result.ok) {
-                error(result.error || `Model ${model} not found`);
+                error(result.error || 'Failed to fetch models');
+                process.exit(1);
+            }
+            const m = result.data.models.find((entry) => entry.id === model || entry.id.endsWith(`/${model}`));
+            if (!m) {
+                error(`Model ${model} not found`);
                 process.exit(1);
             }
-            const m = result.data;
             if (options.json) {
                 console.log(JSON.stringify(m, null, 2));
                 return;
@@ -102,12 +109,18 @@ export function modelsCommands() {
     });
     cmd
         .command('set-default <model>')
-        .description('Set the default model')
+        .description('Set the default model for a provider')
         .option('--json', 'Output as JSON')
         .action(async (model, options) => {
+        const provider = extractProvider(model);
+        if (!provider) {
+            error('Model must include provider prefix (e.g. anthropic/claude-sonnet-4-20250514)');
+            process.exit(1);
+        }
         const spin = spinner(`Setting default model to ${model}...`).start();
         try {
-            const result = await api.post('/api/providers/models/default', { model });
+            const modelName = model.slice(provider.length + 1);
+            const result = await api.post(`/api/chat/providers/${encodeURIComponent(provider)}/configure`, { defaultModel: modelName });
             spin.stop();
             if (!result.ok) {
                 error(result.error || 'Failed to set default model');
@@ -132,7 +145,7 @@ export function modelsCommands() {
         .action(async (options) => {
         const spin = spinner('Fetching aliases...').start();
         try {
-            const result = await api.get('/api/providers/models/aliases');
+            const result = await api.get('/api/chat/models');
             spin.stop();
             if (!result.ok) {
                 error(result.error || 'Failed to fetch aliases');
@@ -143,14 +156,13 @@ export function modelsCommands() {
                 console.log(JSON.stringify(aliases, null, 2));
                 return;
             }
-            const entries = Object.entries(aliases);
-            if (entries.length === 0) {
+            if (!aliases || aliases.length === 0) {
                 info('No model aliases configured.');
                 return;
             }
-            const table = createTable(['Alias', 'Resolves To']);
-            for (const [alias, target] of entries) {
-                table.push([chalk.cyan(alias), target]);
+            const table = createTable(['Alias', 'Provider', 'Model']);
+            for (const entry of aliases) {
+                table.push([chalk.cyan(entry.alias), entry.provider, entry.model]);
             }
             console.log(table.toString());
         }
@@ -162,15 +174,16 @@ export function modelsCommands() {
     });
     cmd
         .command('test <model>')
-        .description('Test a model with a quick prompt')
+        .description('Test a model provider health check')
         .option('--json', 'Output as JSON')
         .action(async (model, options) => {
-        const spin = spinner(`Testing ${model}...`).start();
+        const provider = extractProvider(model) ?? model;
+        const spin = spinner(`Testing ${provider}...`).start();
         try {
-            const result = await api.post(`/api/providers/models/${encodeURIComponent(model)}/test`);
+            const result = await api.post(`/api/chat/providers/${encodeURIComponent(provider)}/health`);
             spin.stop();
             if (!result.ok) {
-                error(result.error || `Failed to test ${model}`);
+                error(result.error || `Failed to test ${provider}`);
                 process.exit(1);
             }
             const data = result.data;
@@ -178,12 +191,12 @@ export function modelsCommands() {
                 console.log(JSON.stringify(data, null, 2));
                 return;
             }
-            if (data.ok) {
+            if (data.healthy) {
                 const latency = data.latencyMs != null ? chalk.dim(` (${data.latencyMs}ms)`) : '';
-                success(`${chalk.cyan(model)} responded successfully${latency}`);
+                success(`${chalk.cyan(provider)} is healthy${latency}`);
             }
             else {
-                error(`${chalk.cyan(model)} test failed: ${data.error || 'Unknown error'}`);
+                error(`${chalk.cyan(provider)} health check failed: ${data.message || 'Unknown error'}`);
             }
         }
         catch (err) {