@praveencs/agent 0.9.29 → 0.10.0

package/README.md

@@ -301,20 +301,66 @@ The daemon auto-discovers scripts and can execute them via the `script.run` tool
 
 ### Plugins
 
-Bundle skills + commands + scripts + hooks into a package:
+Bundle native Node.js tools, skills, commands, scripts, and hooks into a single distributable package. The **Agent Hub** acts as the official registry for community plugins.
 
+```bash
+# Install the official GitHub plugin from the Hub
+agent plugins install github
+
+# Or install from a local path
+agent plugins install ./my-plugin
+
+# List installed plugins
+agent plugins list
 ```
-my-plugin/
-├── plugin.json
-├── skills/
-├── commands/
-├── scripts/
-└── hooks/
+
+**Featured Plugin: GitHub (`github`)**
+- Grants the agent zero-dependency native control over GitHub.
+- Can create repos, open PRs, and manipulate Issues.
+- Unlocks **Advanced Global Search** natively.
+- Can view, dispatch, and monitor **GitHub Actions CI/CD workflows**.
+
+**16 Plugins Available** — Slack, Notion, Vercel, Supabase, Stripe, AWS, Discord, OpenAI, Linear, Docker, MongoDB, Firebase, Telegram, HuggingFace, Resend.
+
+### 🐳 Sandboxed Execution
+
+Run commands safely inside Docker containers:
+
+```bash
+agent sandbox start              # Spin up ephemeral container
+agent sandbox status             # Container info
+agent sandbox stop               # Destroy sandbox
 ```
 
+### 🐝 Multi-Agent Swarm
+
+Coordinate specialized agents (Planner, Coder, Reviewer, Researcher, Tester):
+
 ```bash
-agent plugins install ./my-plugin
-agent plugins list
+agent swarm start "Build a REST API with auth"
+agent swarm status               # View agents & tasks
+agent swarm roles                # List available roles
+```
+
+### 🖥️ Desktop Automation
+
+Cross-platform desktop control (Linux, macOS, Windows):
+
+```bash
+agent desktop screenshot         # Capture screen
+agent desktop click 500 300      # Mouse click
+agent desktop type "Hello" --enter
+agent desktop hotkey ctrl+s      # Keyboard shortcut
+```
+
+### 🌈 Multimodal Interfaces
+
+Voice, vision, and speech powered by OpenAI:
+
+```bash
+agent multimodal transcribe audio.wav      # Whisper STT
+agent multimodal analyze image.png         # GPT-4o Vision
+agent multimodal speak "Done!"             # TTS
 ```
 
 ### Lifecycle Hooks

package/ROADMAP.md

@@ -1,71 +1,63 @@
 # 🛣️ Roadmap: The Future of @praveencs/agent
 
-We have built a robust autonomous agent runtime (`v0.7.x`). But this is just the beginning.
-Here is our vision for the next major milestones.
+We have built a robust autonomous agent runtime (`v0.9.x`). Here is our progress on the major milestones.
 
-## Phase 1: Robustness & Safety (Current Focus)
-- [ ] **Sandboxed Execution**: Run all shell skills inside ephemeral Docker containers to prevent accidental system damage.
-- [ ] **Permission Scopes**: Fine-grained access control (e.g., "Allow read access to `/project` but write access only to `/project/src`").
-- [ ] **Secrets Management**: Secure, encrypted storage for API keys integrated with system keychains.
+## Phase 1: Robustness & Safety ✅
+- [x] **Sandboxed Execution**: Run all shell commands inside ephemeral Docker containers via `src/sandbox/`. Configure image, network, and mounts in `agent.yaml`.
+- [x] **Permission Scopes**: Fine-grained access control via the policy engine (`config.policy`).
+- [x] **Secrets Management**: Encrypted credential vault with `agent studio` Credentials page and `secrets.get/set/list` tools.
 
-## Phase 2: Multi-Agent Collaboration (The Swarm)
-- [ ] **Agent-to-Agent Protocol**: Define a standard schema for agents to send messages and delegate tasks to each other.
-- [ ] **Specialized Personas**:
-    - `Coder Agent`: Writes and tests code.
-    - `Reviewer Agent`: Critiques pull requests.
-    - `Architect Agent`: High-level system design.
-- [ ] **Orchestrator**: A master process that spins up specialized agents for a complex goal.
+## Phase 2: Multi-Agent Collaboration (The Swarm) ✅
+- [x] **Agent-to-Agent Protocol**: Event-driven `MessageBus` for inter-agent communication with pub/sub, broadcast, and history.
+- [x] **Specialized Personas**: 5 built-in roles — Planner, Coder, Reviewer, Researcher, Tester — each with system prompts and tool permissions.
+- [x] **Orchestrator**: `SwarmOrchestrator` spawns agents, assigns tasks, handles delegation chains up to configurable depth.
 
-## Phase 3: Multimodal Interfaces
-- [ ] **Voice Interface**: Speak to your agent ("Deploy this to prod") and hear responses.
-- [ ] **Vision Capabilities**: Allow the agent to "see" your screen or read images (e.g., "Fix the CSS on this screenshot").
+## Phase 3: Multimodal Interfaces ✅
+- [x] **Voice Interface**: Speech-to-text via OpenAI Whisper (`agent multimodal transcribe`).
+- [x] **Vision Capabilities**: Image analysis via GPT-4o vision (`agent multimodal analyze`) — supports local files and URLs.
+- [x] **Text-to-Speech**: Generate spoken audio with 6 voices (`agent multimodal speak`).
 - [ ] **IDE Integration**: VS Code extension to have the agent live in your editor sidebar.
 
 ## Phase 4: The Agent Cloud
-- [ ] **Skill Hub**: A public registry (npm-style) to share and install community skills.
+- [x] **Skill Hub**: Agent Hub with 16 plugins (registry + remote install from GitHub).
 - [ ] **Remote Execution**: Run the heavy agent logic on a cloud server while controlling it from your laptop.
-- [ ] **Web Dashboard**: Real-time visualization of agent thought processes, memory graph, and task plans.
+- [x] **Web Dashboard**: Agent Studio with real-time visualization, 18 panels, and REST API.
 
 ## Phase 5: Plugin Ecosystem & Extensibility ✅
-- [x] **Lifecycle Hooks**: Event-driven hook system (`before:tool`, `after:step`, `before:plan`, etc.) allowing custom scripts to intercept and validate agent execution at every stage.
-- [x] **Lightweight Commands**: Reusable goal templates defined as markdown files with YAML frontmatter — no `skill.json` boilerplate needed. Auto-detected by `agent run <command-name>`.
-- [x] **Multi-CLI Orchestration**: First-class tool wrappers for external AI CLIs (`cli.cursor`, `cli.codex`, `cli.gemini`, `cli.claude`) so the agent can delegate specialized coding sub-tasks to the right tool.
-- [x] **Plugin System**: Distributable bundles (`plugin.json`) that package skills, commands, hooks, tools, and scripts together. Install with `agent plugins install <path>`.
-- [x] **Scripts System**: Repeatable, scriptable tasks defined via `script.yaml` — direct shell/Node execution without LLM involvement. Support for named arguments, environment variables, interpreter auto-detection, timeouts, and confirmation prompts. Works both standalone (`.agent/scripts/`) and inside plugins.
-- [x] **New CLI Commands**: `agent hooks list|add|events`, `agent commands list`, `agent plugins list|install|remove`, `agent scripts list|run|show`.
+- [x] **Lifecycle Hooks**: Event-driven hook system (`before:tool`, `after:step`, `before:plan`, etc.).
+- [x] **Lightweight Commands**: Reusable goal templates as markdown files with YAML frontmatter.
+- [x] **Multi-CLI Orchestration**: Tool wrappers for external AI CLIs (`cli.cursor`, `cli.codex`, `cli.gemini`, `cli.claude`).
+- [x] **Plugin System**: 16 distributable plugins (GitHub, Slack, Notion, Vercel, Supabase, Stripe, AWS, Discord, OpenAI, Linear, Docker, MongoDB, Firebase, Telegram, HuggingFace, Resend).
+- [x] **Scripts System**: Repeatable scriptable tasks via `script.yaml`.
+- [x] **New CLI Commands**: `agent hooks|commands|plugins|scripts|sandbox|swarm|desktop|multimodal`.
 
 ## Phase 6: Interactive CLI Experience ✅
-- [x] **Interactive REPL**: When the user types `agent` with no arguments, launch a conversational session with multi-turn context, slash commands, and inline tool execution.
-- [x] **Slash Commands**: Built-in `/help`, `/skills`, `/commands`, `/hooks`, `/model`, `/compact`, `/clear`, `/exit` — plus any user-defined commands are auto-registered as `/command-name`.
-- [x] **Rich Terminal UI**: Bordered welcome banner, ora spinners during LLM thinking, inline tool call badges (⚡ running → ✓/✗), completion summaries with timing.
-- [x] **Tab Completion**: Autocomplete slash commands and user commands in the REPL.
-- [x] **Conversation Context**: Multi-turn session with conversation compaction support.
+- [x] **Interactive REPL**: Conversational sessions with multi-turn context and slash commands.
+- [x] **Slash Commands**: Built-in `/help`, `/skills`, `/commands`, `/hooks`, `/model`, `/compact`, `/clear`, `/exit`.
+- [x] **Rich Terminal UI**: Bordered welcome banner, spinners, inline tool call badges, completion timing.
+- [x] **Tab Completion**: Autocomplete slash commands in the REPL.
+- [x] **Conversation Context**: Multi-turn with conversation compaction support.
 
-## Phase 7: Desktop Automation (All OS)
-Cross-platform desktop control so the agent can operate any application — not just the terminal.
+## Phase 7: Desktop Automation (All OS) ✅
+- [x] **`desktop.screenshot`**: Cross-platform capture (scrot/gnome-screenshot, screencapture, PowerShell).
+- [x] **`desktop.click`** / **`desktop.doubleClick`** / **`desktop.rightClick`**: Mouse control via xdotool/osascript.
+- [x] **`desktop.type`** / **`desktop.hotkey`**: Keyboard input and shortcut combinations.
+- [x] **`desktop.scroll`**: Scroll at position.
+- [x] **`desktop.drag`**: Drag from point A to point B.
+- [x] **Cross-platform engine**: Works on Linux (xdotool), macOS (osascript/cliclick), and Windows (PowerShell).
 
-### Layer 1: Screen Control (`desktop.*` tools)
-- [ ] **`desktop.screenshot`**: Capture the screen or a specific window, return as base64 for vision LLMs.
-- [ ] **`desktop.click`** / **`desktop.doubleClick`** / **`desktop.rightClick`**: Mouse control at pixel coordinates.
-- [ ] **`desktop.type`** / **`desktop.hotkey`**: Keyboard input and shortcut combinations (e.g., `Ctrl+S`).
-- [ ] **`desktop.scroll`**: Scroll at position or within active window.
-- [ ] **`desktop.drag`**: Drag from point A to point B.
-- [ ] **Cross-platform engine**: Built on [`nut.js`](https://nutjs.dev) — works on Windows, macOS, and Linux with a single API.
+### Layer 2: AI-Powered Vision ✅
+- [x] **Vision Loop**: Screenshot → GPT-4o analysis → action → verify cycle via MultimodalEngine.
+- [ ] **Element Detection**: Use image template matching for UI element recognition.
+- [ ] **OCR Integration**: Extract text from screen regions.
 
-### Layer 2: AI-Powered Vision
-- [ ] **Anthropic Computer Use**: Send screenshots to Claude, receive structured actions (`click`, `type`, `scroll`) with coordinates.
-- [ ] **Vision Loop**: `screenshot → LLM analysis → action → screenshot → verify` cycle for goal-driven desktop automation.
-- [ ] **Element Detection**: Use image template matching (`@nut-tree/template-matcher`) to find UI elements by visual appearance.
-- [ ] **OCR Integration**: Extract text from screen regions for reading non-accessible UI content.
-
-### Layer 3: OS-Native Accessibility APIs
-- [ ] **macOS**: AppleScript / JXA for native app control (`tell application "Safari" to open location`).
-- [ ] **Windows**: PowerShell + UI Automation API for Win32/WPF/UWP app control.
-- [ ] **Linux**: `xdotool` + AT-SPI2 for X11/Wayland window management and GTK/Qt app inspection.
-- [ ] **Window Management**: `desktop.focus`, `desktop.resize`, `desktop.list_windows` via `active-win` / `node-window-manager`.
+### Layer 3: OS-Native Accessibility APIs (Partial)
+- [x] **macOS**: AppleScript/osascript for native app control.
+- [x] **Linux**: xdotool for X11 window management.
+- [ ] **Windows**: PowerShell + UI Automation API.
 
 ### Layer 4: Browser Automation
-- [ ] **Playwright Integration**: `desktop.browser.open`, `desktop.browser.click`, `desktop.browser.fill` for web app automation.
+- [ ] **Playwright Integration**: `desktop.browser.open`, `desktop.browser.click`, `desktop.browser.fill`.
 - [ ] **Cookie/Session Persistence**: Maintain browser sessions across agent runs.
 
 ### Use Cases

package/agent-skills/README.md

@@ -0,0 +1,192 @@
+<p align="center">
+  <h1 align="center">📦 Agent Hub</h1>
+  <p align="center">
+    The official Plugin and Skill registry for <a href="https://github.com/praveencs87/agent"><b>@praveencs/agent</b></a><br/>
+    Community-built plugins and skills for AI-powered automation
+  </p>
+  <p align="center">
+    <a href="https://github.com/praveencs87/agent-skills"><img src="https://img.shields.io/badge/skills-16-blueviolet" alt="Skills"></a>
+    <a href="https://github.com/praveencs87/agent"><img src="https://img.shields.io/badge/agent-v0.1.0-blue" alt="Agent"></a>
+    <a href="https://github.com/praveencs87/agent-skills/blob/main/LICENSE"><img src="https://img.shields.io/badge/license-MIT-green" alt="License"></a>
+  </p>
+</p>
+
+---
+
+## Installation
+
+### Installing Plugins
+```bash
+agent plugins install <plugin-name>
+# Example: 
+agent plugins install github
+```
+
+### Installing Skills
+```bash
+agent skills install <skill-name>
+```
+
+### Manual Installation
+```bash
+git clone https://github.com/praveencs87/agent-skills.git
+cp -r agent-skills/skills/<skill-name> .agent/skills/
+```
+
+---
+
+## 📋 Table of Contents
+
+- [Plugins](#plugins)
+- [Coding & IDEs](#coding--ides)
+- [Git & Version Control](#git--version-control)
+- [DevOps & Cloud](#devops--cloud)
+- [Communication](#communication)
+- [Productivity](#productivity)
+- [Search & Research](#search--research)
+- [CLI Utilities](#cli-utilities)
+- [Data & Analytics](#data--analytics)
+- [Browser & Automation](#browser--automation)
+
+---
+
+## Plugins
+
+Plugins are full-capability bundles that provide native Node.js tools, hooks, scripts, and skills in a single standalone package.
+
+| Plugin | Description | Capabilities |
+|--------|-------------|--------------|
+| [github](plugins/github/) | Manage repos, PRs, issues, and CI/CD workflows on GitHub. | `tools/github.js`, `skills/github-manager.md` |
+| [slack](plugins/slack/) | Send messages, manage channels, search history, and upload files to Slack. | `tools/slack.js`, `skills/slack-manager.md` |
+| [notion](plugins/notion/) | Create/manage Notion pages, databases, blocks, and search workspace. | `tools/notion.js`, `skills/notion-manager.md` |
+| [vercel](plugins/vercel/) | Deploy projects, manage domains, and configure env vars on Vercel. | `tools/vercel.js`, `skills/vercel-manager.md` |
+| [supabase](plugins/supabase/) | Query Postgres, manage auth users, and use storage on Supabase. | `tools/supabase.js`, `skills/supabase-manager.md` |
+| [stripe](plugins/stripe/) | Manage customers, subscriptions, checkout sessions, and refunds on Stripe. | `tools/stripe.js`, `skills/stripe-manager.md` |
+| [aws](plugins/aws/) | Manage S3 buckets, Lambda functions, EC2 instances, and CloudWatch logs. | `tools/aws.js`, `skills/aws-manager.md` |
+| [discord](plugins/discord/) | Send messages, manage channels and roles, use webhooks on Discord. | `tools/discord.js`, `skills/discord-manager.md` |
+| [openai](plugins/openai/) | Direct GPT chat, DALL-E image gen, Whisper transcription, TTS, and embeddings. | `tools/openai.js`, `skills/openai-manager.md` |
+| [linear](plugins/linear/) | Manage issues, projects, teams, and cycles on Linear. | `tools/linear.js`, `skills/linear-manager.md` |
+| [docker](plugins/docker/) | Build images, run containers, manage compose stacks on Docker. | `tools/docker.js`, `skills/docker-manager.md` |
+| [mongodb](plugins/mongodb/) | Query, insert, update, and aggregate data on MongoDB Atlas. | `tools/mongodb.js`, `skills/mongodb-manager.md` |
+| [firebase](plugins/firebase/) | Manage Firestore docs, Auth users, and Cloud Storage on Firebase. | `tools/firebase.js`, `skills/firebase-manager.md` |
+| [telegram](plugins/telegram/) | Send messages, photos, polls, and manage groups via Telegram bots. | `tools/telegram.js`, `skills/telegram-manager.md` |
+| [huggingface](plugins/huggingface/) | Run ML inference and search the Hugging Face model hub. | `tools/huggingface.js`, `skills/huggingface-manager.md` |
+| [resend](plugins/resend/) | Send transactional emails and manage domains via Resend. | `tools/resend.js`, `skills/resend-manager.md` |
+
+---
+
+## Coding & IDEs
+
+| Skill | Description | Tools |
+|-------|-------------|-------|
+| [open-vscode](skills/open-vscode/) | Opens VS Code in the current or specified directory | `cmd.run` |
+| [project-scaffold](skills/project-scaffold/) | Scaffold new projects (Node.js, Python, React, Next.js, Express) | `cmd.run`, `fs.write` |
+| [npm-publish](skills/npm-publish/) | Prepare and publish npm packages with version bumping | `cmd.run`, `fs.read`, `fs.write` |
+| [code-review](skills/code-review/) | Review code changes for bugs, security issues, and best practices | `cmd.run`, `git.diff`, `fs.read` |
+
+## Git & Version Control
+
+| Skill | Description | Tools |
+|-------|-------------|-------|
+| [git-commit](skills/git-commit/) | Stages changes and creates conventional git commits | `cmd.run`, `git.status`, `git.diff` |
+
+## DevOps & Cloud
+
+| Skill | Description | Tools |
+|-------|-------------|-------|
+| [docker-deploy](skills/docker-deploy/) | Build, run, and manage Docker containers and compose stacks | `cmd.run` |
+| [system-monitor](skills/system-monitor/) | Monitor system resources — CPU, memory, disk, processes | `cmd.run` |
+| [log-analyzer](skills/log-analyzer/) | Analyze log files — find errors, patterns, and anomalies | `cmd.run`, `fs.read` |
+
+## Communication
+
+| Skill | Description | Tools |
+|-------|-------------|-------|
+| [send-email](skills/send-email/) | Sends emails using nodemailer via SMTP | `cmd.run` |
+
+## Productivity
+
+| Skill | Description | Tools |
+|-------|-------------|-------|
+| [create-note](skills/create-note/) | Creates formatted markdown notes and saves them as files | `fs.write` |
+
+## Search & Research
+
+| Skill | Description | Tools |
+|-------|-------------|-------|
+| [web-search](skills/web-search/) | Search the web using curl and extract relevant information | `cmd.run` |
+
+## CLI Utilities
+
+| Skill | Description | Tools |
+|-------|-------------|-------|
+| [file-organizer](skills/file-organizer/) | Organizes files by type, date, or custom rules | `cmd.run`, `fs.read`, `fs.write` |
+| [backup](skills/backup/) | Backup files and directories using tar, rsync, or zip | `cmd.run` |
+| [cron-scheduler](skills/cron-scheduler/) | Manage cron jobs — create, list, edit, delete | `cmd.run` |
+
+## Data & Analytics
+
+| Skill | Description | Tools |
+|-------|-------------|-------|
+| [db-query](skills/db-query/) | Query databases — SQLite, PostgreSQL, MySQL via CLI | `cmd.run` |
+
+## Browser & Automation
+
+| Skill | Description | Tools |
+|-------|-------------|-------|
+| [api-tester](skills/api-tester/) | Test REST APIs with curl — GET, POST, PUT, DELETE | `cmd.run` |
+
+---
+
+## 🛠️ Creating a Skill
+
+Each skill is a folder containing:
+
+```
+my-skill/
+├── skill.json     # Manifest (name, description, tools, permissions)
+└── prompt.md      # Instructions for the AI agent
+```
+
+### `skill.json` Example
+```json
+{
+  "name": "my-skill",
+  "version": "1.0.0",
+  "description": "What this skill does",
+  "tools": ["cmd.run"],
+  "permissions": {
+    "required": ["exec"]
+  },
+  "entrypoint": "prompt.md",
+  "state": "approved"
+}
+```
+
+### `prompt.md` Example
+```markdown
+Describe what the agent should do when this skill is invoked.
+Include specific commands, parameters, and expected outputs.
+```
+
+---
+
+## 🤝 Contributing
+
+1. Fork this repository
+2. Create your skill folder under `skills/`
+3. Add `skill.json` and `prompt.md`
+4. Add your skill to `registry.json`
+5. Submit a Pull Request
+
+### Guidelines
+- Skills should be focused and single-purpose
+- Include clear instructions in `prompt.md`
+- Specify minimum required permissions
+- Test your skill with `agent run --skill <name> "<goal>"`
+
+---
+
+## License
+
+MIT © [praveencs](https://github.com/praveencs87)

package/agent-skills/plugins/aws/plugin.json

@@ -0,0 +1,22 @@
+{
+    "name": "aws",
+    "version": "1.0.0",
+    "description": "Manage S3 buckets, Lambda functions, EC2 instances, and CloudWatch logs on AWS.",
+    "author": "Praveen",
+    "category": "DevOps & Cloud",
+    "tags": [
+        "aws",
+        "s3",
+        "lambda",
+        "ec2",
+        "cloud"
+    ],
+    "exports": {
+        "tools": [
+            "tools/aws.js"
+        ],
+        "skills": [
+            "skills/aws-manager.md"
+        ]
+    }
+}

package/agent-skills/plugins/aws/skills/aws-manager.md

@@ -0,0 +1,17 @@
+# AWS Manager
+
+## Description
+This skill enables you to manage S3 buckets, Lambda functions, EC2 instances, and CloudWatch logs on AWS.
+
+## Instructions
+1. This plugin uses the **AWS CLI** (`aws`). It must be installed and configured on the host machine.
+2. For S3 operations: `s3_list_buckets` shows all buckets, `s3_list_objects` lists files in a bucket with optional `prefix` filtering.
+3. To upload a local file to S3, use `action: "s3_upload"` with `file` (local path), `bucket`, and `key` (destination path).
+4. For Lambda, use `lambda_invoke` with `function_name` and a `payload` object. The response contains the function's return value.
+5. For EC2, use `ec2_list` to see instances (filtered by `state`: running, stopped). Use `ec2_start` / `ec2_stop` with the `instance_id`.
+6. For CloudWatch, first `cloudwatch_log_groups` to discover groups, then `cloudwatch_logs` with `log_group` and `log_stream` to fetch entries.
+7. Set `region` parameter to target a specific AWS region, or it defaults to `AWS_DEFAULT_REGION` or `us-east-1`.
+8. Credentials can be configured via `AWS_ACCESS_KEY_ID` + `AWS_SECRET_ACCESS_KEY` or `AWS_PROFILE`.
+
+## Input Variables
+{{input}}

package/agent-skills/plugins/aws/tools/aws.js

@@ -0,0 +1,152 @@
+/**
+ * AWS automation tool.
+ * Provides S3, Lambda, EC2, and CloudWatch management via AWS CLI commands.
+ * Uses the AWS CLI (`aws`) which must be installed and configured on the host.
+ */
+import { execFile } from 'node:child_process';
+import { promisify } from 'node:util';
+
+const execFileAsync = promisify(execFile);
+
+export async function execute(params) {
+    const { action, ...args } = params;
+
+    // Verify AWS credentials are available
+    if (!process.env.AWS_ACCESS_KEY_ID && !process.env.AWS_PROFILE) {
+        throw new Error('AWS credentials not configured. Export AWS_ACCESS_KEY_ID + AWS_SECRET_ACCESS_KEY, or set AWS_PROFILE.');
+    }
+
+    const region = args.region || process.env.AWS_DEFAULT_REGION || 'us-east-1';
+
+    /** Run an AWS CLI command and return parsed JSON */
+    async function awsCli(service, command, cliArgs = []) {
+        const fullArgs = [service, command, '--region', region, '--output', 'json', ...cliArgs];
+        try {
+            const { stdout } = await execFileAsync('aws', fullArgs, {
+                timeout: 30000,
+                maxBuffer: 10 * 1024 * 1024,
+                env: process.env
+            });
+            try {
+                return JSON.parse(stdout);
+            } catch {
+                return { output: stdout.trim() };
+            }
+        } catch (err) {
+            throw new Error(`AWS CLI Error: ${err.stderr || err.message}`);
+        }
+    }
+
+    try {
+        switch (action) {
+            // --- S3 ---
+            case 's3_list_buckets':
+                return await awsCli('s3api', 'list-buckets');
+
+            case 's3_list_objects':
+                return await awsCli('s3api', 'list-objects-v2', [
+                    '--bucket', args.bucket,
+                    '--max-items', String(args.limit || 50),
+                    ...(args.prefix ? ['--prefix', args.prefix] : [])
+                ]);
+
+            case 's3_create_bucket':
+                return await awsCli('s3api', 'create-bucket', [
+                    '--bucket', args.bucket,
+                    ...(region !== 'us-east-1' ? ['--create-bucket-configuration', `LocationConstraint=${region}`] : [])
+                ]);
+
+            case 's3_delete_bucket':
+                return await awsCli('s3api', 'delete-bucket', ['--bucket', args.bucket]);
+
+            case 's3_upload': {
+                const dest = `s3://${args.bucket}/${args.key}`;
+                return await awsCli('s3', 'cp', [args.file, dest]);
+            }
+
+            case 's3_download': {
+                const src = `s3://${args.bucket}/${args.key}`;
+                return await awsCli('s3', 'cp', [src, args.file || `./${args.key}`]);
+            }
+
+            // --- LAMBDA ---
+            case 'lambda_list':
+                return await awsCli('lambda', 'list-functions', ['--max-items', String(args.limit || 20)]);
+
+            case 'lambda_invoke':
+                return await awsCli('lambda', 'invoke', [
+                    '--function-name', args.function_name,
+                    '--payload', JSON.stringify(args.payload || {}),
+                    '--cli-binary-format', 'raw-in-base64-out',
+                    '/dev/stdout'
+                ]);
+
+            case 'lambda_get':
+                return await awsCli('lambda', 'get-function', ['--function-name', args.function_name]);
+
+            // --- EC2 ---
+            case 'ec2_list':
+                return await awsCli('ec2', 'describe-instances', [
+                    '--filters', `Name=instance-state-name,Values=${args.state || 'running'}`,
+                    '--max-items', String(args.limit || 20)
+                ]);
+
+            case 'ec2_start':
+                return await awsCli('ec2', 'start-instances', ['--instance-ids', args.instance_id]);
+
+            case 'ec2_stop':
+                return await awsCli('ec2', 'stop-instances', ['--instance-ids', args.instance_id]);
+
+            // --- CLOUDWATCH ---
+            case 'cloudwatch_logs':
+                return await awsCli('logs', 'get-log-events', [
+                    '--log-group-name', args.log_group,
+                    '--log-stream-name', args.log_stream,
+                    '--limit', String(args.limit || 50)
+                ]);
+
+            case 'cloudwatch_log_groups':
+                return await awsCli('logs', 'describe-log-groups', [
+                    '--limit', String(args.limit || 20)
+                ]);
+
+            default:
+                throw new Error(`Unknown action: ${action}`);
+        }
+    } catch (err) {
+        return { error: err.message, stack: err.stack };
+    }
+}
+
+export const definition = {
+    name: 'aws',
+    description: 'Manage AWS resources: S3 buckets, Lambda functions, EC2 instances, and CloudWatch logs. Requires AWS CLI installed and credentials configured.',
+    parameters: {
+        type: 'object',
+        required: ['action'],
+        properties: {
+            action: {
+                type: 'string',
+                description: 'The AWS operation to perform',
+                enum: [
+                    's3_list_buckets', 's3_list_objects', 's3_create_bucket', 's3_delete_bucket', 's3_upload', 's3_download',
+                    'lambda_list', 'lambda_invoke', 'lambda_get',
+                    'ec2_list', 'ec2_start', 'ec2_stop',
+                    'cloudwatch_logs', 'cloudwatch_log_groups'
+                ]
+            },
+            bucket: { type: 'string', description: 'S3 bucket name' },
+            key: { type: 'string', description: 'S3 object key (path within bucket)' },
+            prefix: { type: 'string', description: 'S3 key prefix for filtering' },
+            file: { type: 'string', description: 'Local file path (for upload/download)' },
+            function_name: { type: 'string', description: 'Lambda function name or ARN' },
+            payload: { type: 'object', description: 'JSON payload for Lambda invocation' },
+            instance_id: { type: 'string', description: 'EC2 instance ID (i-xxx)' },
+            state: { type: 'string', description: 'EC2 instance state filter (running, stopped, etc.)' },
+            log_group: { type: 'string', description: 'CloudWatch log group name' },
+            log_stream: { type: 'string', description: 'CloudWatch log stream name' },
+            region: { type: 'string', description: 'AWS region (default: us-east-1)' },
+            limit: { type: 'number', description: 'Max results to return' }
+        }
+    }
+};

package/agent-skills/plugins/discord/plugin.json

@@ -0,0 +1,22 @@
+{
+    "name": "discord",
+    "version": "1.0.0",
+    "description": "Send messages, manage channels and roles, and interact with Discord servers via webhooks and bot API.",
+    "author": "Praveen",
+    "category": "Communication",
+    "tags": [
+        "discord",
+        "messaging",
+        "chat",
+        "bot",
+        "community"
+    ],
+    "exports": {
+        "tools": [
+            "tools/discord.js"
+        ],
+        "skills": [
+            "skills/discord-manager.md"
+        ]
+    }
+}

package/agent-skills/plugins/discord/skills/discord-manager.md

@@ -0,0 +1,15 @@
+# Discord Manager
+
+## Description
+This skill enables you to send messages, manage channels and roles, and interact with Discord servers.
+
+## Instructions
+1. Discord uses **snowflake IDs** (large numbers like `1234567890123456789`) for channels, guilds, users, and roles. You must provide numeric IDs, not names.
+2. For quick notifications without a bot token, use `action: "send_webhook"` with a `webhook_url`. Webhooks only need a URL, no authentication.
+3. To send a rich message, use `embeds` (an array of embed objects with `title`, `description`, `color`, `fields`, etc.).
+4. To manage a server, you need the `guild_id`. Use `action: "get_guild"` to verify access.
+5. Channel types: `0` = text channel, `2` = voice channel, `4` = category. Use `parent_id` to nest a channel under a category.
+6. If the tool returns an authentication error, the user needs to export `DISCORD_BOT_TOKEN`. Create a bot at https://discord.com/developers/applications.
+
+## Input Variables
+{{input}}