@poolzin/pool-bot 2026.3.7 → 2026.3.10

package/CHANGELOG.md

@@ -1,3 +1,43 @@
+## v2026.3.10 (2026-03-10)
+
+### Features
+- **Telemetry Alert System:** intelligent alerting engine for proactive monitoring
+  - Real-time alert evaluation with configurable thresholds
+  - Built-in rules for cron failures, security scans, swarm tasks, gateway health
+  - Webhook and announce channel notifications
+  - Alert history tracking with automatic cleanup
+- **Swarm State Persistence:** swarm task state now persisted to `~/.poolbot/swarm-state.json`
+  - Survives gateway restarts without losing task queue
+  - Automatic state recovery on startup
+- **HexStrike Bridge Telemetry:** comprehensive metrics for HexStrike AI integration
+  - Request latency tracking with histograms
+  - Rate limiting with configurable thresholds
+  - Health check monitoring with automatic failover
+- **Critical Bug Fix:** telemetry service now properly starts with gateway (was missing `await`)
+
+### Improvements
+- Optimized alert thresholds for production workloads
+- Enhanced gateway instrumentation with memory and CPU metrics
+- Improved error handling in telemetry pipeline
+
+---
+
+## v2026.3.9 (2026-03-09)
+
+### Features
+- **Modular Skills System:** new comprehensive skills management system in `src/skills/` — 8 core modules providing SKILL.md parsing, security scanning, registry management, progressive disclosure loading, and CLI commands (`poolbot mods`)
+  - `types.ts`: Type definitions compatible with agentskills.io specification
+  - `parser.ts`: YAML frontmatter parser for SKILL.md files with validation
+  - `registry.ts`: EventEmitter-based skills registry with lifecycle management
+  - `loader.ts`: Progressive disclosure loader with dependency resolution
+  - `security.ts`: Security scanner for prompt injection and path traversal detection
+  - `commands.ts`: CLI commands for skill management
+  - `index.ts`: Public API exports
+  - Comprehensive test suite with 52 passing tests
+- **Integration Plans:** detailed implementation plans for external project integrations (Page Agent, HexStrike AI, xyOps) via Gateway Node protocol
+
+---
+
 ## v2026.3.6 (2026-03-06)
 
 ### Features

package/README.md

@@ -1,123 +1,201 @@
 # 🎱 Pool Bot
 
-> Your AI pool of knowledge
+> **The AI Gateway for Teams That Need More Than Chatbots**
 
-**Pool Bot** is a professional AI assistant with advanced PLCODE integrations and development-focused features.
+Pool Bot transforms your messaging apps into a powerful AI command center. Connect Claude, GPT, Gemini, and local models to WhatsApp, Telegram, Discord, Slack, Signal, and 15+ other channels—then extend with custom skills for your specific workflows.
+
+[![npm version](https://img.shields.io/npm/v/@poolzin/pool-bot)](https://www.npmjs.com/package/@poolzin/pool-bot)
+[![License: MIT](https://img.shields.io/badge/License-MIT-blue.svg)](LICENSE)
+[![Documentation](https://img.shields.io/badge/docs-docs.poolbot.dev-green)](https://docs.poolbot.dev)
 
 ![Pool Bot](README-header.png)
 
-## 🌟 What Makes Pool Bot Different
+---
 
-- 🎱 **PLCODE Integration** - Native support for coding workflows
-- 🧠 **Enhanced Memory** - Advanced context management
-- 🛠️ **Developer Tools** - Built for development teams
-- 🔒 **Private & Secure** - Your data, your control
-- 🌐 **Multi-Channel Support** - WhatsApp, Telegram, Discord, Slack, Signal, and more
+## ✨ What Makes Pool Bot Different
 
-## Installation
+| Feature | What It Means For You |
+|---------|----------------------|
+| **🧩 Modular Skills System** | Add capabilities via YAML-based skills—no coding required for most extensions |
+| **🌐 Universal Gateway** | One AI backend, unlimited channels. Add new messaging platforms without reconfiguration |
+| **🧠 Advanced Memory** | MMR reranking, temporal decay, hybrid search—conversations that actually remember context |
+| **⚡ Provider Infrastructure** | Multi-key pools, intelligent rate limiting, automatic failover across 10+ LLM providers |
+| **🔒 Security-First** | Command approval workflows, audit logging, PII detection, configurable guardrails |
+| **📱 Device Control** | Pair iOS/Android devices for remote automation, screenshots, and notifications |
 
-```bash
-npm install -g @poolzin/pool-bot
-poolbot --help
-```
+---
 
-## Quick Start
+## 🚀 Quick Start
 
 ```bash
+# Install globally
+npm install -g @poolzin/pool-bot
+
 # Start the gateway
 poolbot gateway start
 
-# Connect WhatsApp
-poolbot channels login
+# Connect your first channel
+poolbot channels login whatsapp
 
-# Check status
+# Check everything is running
 poolbot status
 ```
 
-## Features
+**That's it.** Your AI assistant is now available on WhatsApp. Add more channels anytime.
 
-- 🌐 **Multi-Channel Support** - WhatsApp, Telegram, Discord, Slack, Signal, and more
-- 🤖 **AI Agents** - Claude, GPT, Gemini, and local models
-- 🧠 **Memory System** - Context-aware conversations with semantic search
-- 📱 **Node Pairing** - Control iOS/Android devices remotely
-- 🎨 **Canvas** - Visual browser automation
-- 🔧 **Skills System** - Extensible plugin architecture
-- 📊 **PLCODE Dashboard** - Real-time development metrics
+---
+
+## 📡 Supported Channels
 
-## Documentation
+**Messaging Platforms:**
+WhatsApp • Telegram • Discord • Slack • Signal • iMessage • Facebook Messenger • Instagram • LINE • Matrix • Mattermost • Zalo • Microsoft Teams
 
-Full documentation is available at [docs.poolbot.dev](https://docs.poolbot.dev)
+**Voice & Web:**
+Web Interface • WebRTC • Voice Calls
 
-## Skills
+**Custom:**
+Webhooks • API • Custom integrations via skills
 
-Pool Bot supports an extensible skills system:
+---
 
-- 🎮 **WebGPU Three.js TSL** - 3D development with WebGPU
-- 🔍 **GitHub Integration** - Issues, PRs, and CI management
-- 🌦️ **Weather** - Current weather and forecasts
-- 📧 **Notion** - Database and page management
-- 🐦 **X/Twitter** - Social media engagement
-- And more...
+## 🤖 Supported AI Providers
 
-See [skills directory](skills/) for the full list.
+| Provider | Models | Features |
+|----------|--------|----------|
+| **Anthropic** | Claude 3.5/3 Opus/Sonnet/Haiku | Full tool use, vision, extended thinking |
+| **OpenAI** | GPT-4o, GPT-4, GPT-3.5 | Function calling, vision, DALL-E |
+| **Google** | Gemini 1.5 Pro/Flash | Multimodal, 1M+ context window |
+| **Groq** | Llama 3, Mixtral | Ultra-fast inference |
+| **Ollama** | Local models | Privacy-first, offline capable |
+| **Together** | 50+ open models | Mixtral, Llama, Qwen, and more |
+| **OpenRouter** | 100+ models | Unified API for multiple providers |
+| **Mistral** | Mistral Large/Medium/Small | European AI, GDPR compliant |
+| **Cohere** | Command R/R+ | Enterprise-focused |
+| **AI21** | Jamba/Jurassic | Long context specialist |
 
-## Companion Apps
+---
 
-Official companion apps for connecting devices to your Pool Bot gateway:
+## 🛠️ Skills System
 
-- **[Pool Apps](https://github.com/plcunha/pool-apps)** - Windows, Android, iOS nodes
-  - 🪟 **Windows Node v2** (Python) - Local LLM, STT/TTS, automation
-  - 📱 **Android** - Native mobile app
-  - 🍎 **iOS/macOS** - Apple ecosystem apps
+Pool Bot's unique skill system lets you add capabilities without touching core code:
 
-## Configuration
+```yaml
+# Example: Weather Skill (skills/weather/SKILL.md)
+---
+name: weather
+description: Get weather forecasts
+tools:
+  - name: get_weather
+    description: Get current weather for a location
+    parameters:
+      location: string
+      units: enum[celsius, fahrenheit]
+---
+```
 
-Pool Bot uses a profile-based configuration system:
+**Built-in Skills:**
+- 🎮 **WebGPU Three.js** - 3D development and visualization
+- 🔍 **GitHub** - Issues, PRs, CI/CD management
+- 🌦️ **Weather** - Current conditions and forecasts
+- 📝 **Notion** - Database and page management
+- 🐦 **X/Twitter** - Social media automation
+- 📧 **Gmail** - Email management
+- 📅 **Calendar** - Schedule management
+- 🔐 **1Password** - Secure credential access
+- 🎨 **Canvas** - Visual browser automation
+- 🖼️ **Image Generation** - DALL-E, Stable Diffusion integration
 
-```bash
-# Use default profile
-poolbot gateway start
+**Create your own:** See [Creating Skills](https://docs.poolbot.dev/skills/creating)
 
-# Use custom profile
-poolbot --profile production gateway start
+---
 
-# Use dev profile (isolated state)
-poolbot --dev gateway start
-```
+## 🏗️ Architecture Highlights
+
+### Provider Infrastructure
+- **Multi-key pools** - Rotate across multiple API keys for higher rate limits
+- **Intelligent routing** - Route by model capability, cost, or latency
+- **Automatic failover** - Seamless fallback when providers degrade
+- **Usage tracking** - Per-key, per-model cost analytics
+
+### Memory System
+- **Semantic search** - Find relevant context across conversation history
+- **MMR reranking** - Diverse, non-redundant context retrieval
+- **Temporal decay** - Recent conversations weighted higher
+- **Hybrid search** - Combine vector similarity with keyword matching
+
+---
 
-## Development
+## 📱 Companion Apps
+
+Extend Pool Bot to your devices:
+
+| Platform | Features | Repository |
+|----------|----------|------------|
+| **Windows** | Local LLM, STT/TTS, automation, system control | [pool-apps](https://github.com/plcunha/pool-apps) |
+| **Android** | Native messaging, notifications, device automation | [pool-apps](https://github.com/plcunha/pool-apps) |
+| **iOS/macOS** | Apple ecosystem integration, Shortcuts support | [pool-apps](https://github.com/plcunha/pool-apps) |
+
+---
+
+## 💻 Development
 
 ```bash
-# Clone repository
+# Clone and setup
 git clone https://github.com/plcunha/pool-bot.git
 cd pool-bot
-
-# Install dependencies
 pnpm install
 
+# Development mode
+pnpm dev
+
 # Build
 pnpm build
 
 # Run tests
 pnpm test
 
-# Start gateway
-node dist/entry.js gateway
+# Lint and format
+pnpm lint
+pnpm format
 ```
 
-## Support
+---
+
+## 📚 Documentation
+
+- **[Getting Started](https://docs.poolbot.dev/getting-started)** - Installation and first steps
+- **[Configuration](https://docs.poolbot.dev/configuration)** - Profiles, channels, and providers
+- **[Skills](https://docs.poolbot.dev/skills)** - Built-in and custom skills
+- **[Architecture](https://docs.poolbot.dev/architecture)** - Mesh networking, memory, security
+- **[API Reference](https://docs.poolbot.dev/api)** - REST API and webhooks
+- **[Deployment](https://docs.poolbot.dev/deployment)** - Docker, cloud, and self-hosting
+
+---
+
+## 🔐 Security
+
+- **Command Approval** - Require human approval for destructive operations
+- **Audit Logging** - Complete trace of all AI actions
+- **PII Detection** - Automatic redaction of sensitive data
+- **Sandboxed Execution** - Isolated environment for untrusted skills
+- **End-to-End Encryption** - For supported channels
+
+See [Security Documentation](https://docs.poolbot.dev/security)
+
+---
+
+## 🤝 Community
+
+- **Discord:** [discord.gg/poolbot](https://discord.gg/poolbot)
+- **GitHub Issues:** [Report bugs, request features](https://github.com/plcunha/pool-bot/issues)
+- **Discussions:** [Q&A, show and tell](https://github.com/plcunha/pool-bot/discussions)
 
-- **Documentation**: https://docs.poolbot.dev
-- **GitHub Issues**: https://github.com/plcunha/pool-bot/issues
-- **Community**: https://discord.com/discord.gg/poolbot
+---
 
-## Author
+## 📄 License
 
-João Vitor Cunha dos Santos
-- Email: jvsantos.cunha@gmail.com
-- GitHub: @plcunha
-- Telegram: @poolzin
+MIT License - see [LICENSE](LICENSE) file for details.
 
-## License
+---
 
-MIT License - see LICENSE file for details
+**Made with 🎱 by [João Vitor Cunha](https://github.com/plcunha)**

package/dist/.buildstamp

@@ -1 +1 @@
-1772591706886
+1773140740628

package/dist/agents/error-classifier.js

@@ -0,0 +1,251 @@
+/**
+ * Error Classification System
+ *
+ * Provides comprehensive error classification for LLM/AI operations.
+ * Detects specific error types like context overflow, rate limits, etc.
+ */
+// OpenAI error patterns
+const OPENAI_PATTERNS = {
+    context_overflow: [
+        /context length exceeded/i,
+        /maximum context length/i,
+        /token limit exceeded/i,
+        /too many tokens/i,
+        /rate_limit_exceeded.*context/i,
+    ],
+    rate_limit: [/rate limit exceeded/i, /too many requests/i, /ratelimit/i],
+    authentication: [/invalid api key/i, /incorrect api key/i, /authentication/i, /unauthorized/i],
+    invalid_request: [/invalid_request_error/i, /bad request/i, /invalid parameter/i],
+};
+// Anthropic error patterns
+const ANTHROPIC_PATTERNS = {
+    context_overflow: [/context window exceeded/i, /maximum token count/i, /too many tokens/i],
+    rate_limit: [/rate limit/i, /too many requests/i],
+    authentication: [/invalid api key/i, /authentication failed/i],
+    authorization: [
+        /forbidden/i,
+        /access denied/i,
+        /unauthorized.*resource/i,
+        /insufficient.*permission/i,
+    ],
+};
+// Google/Gemini error patterns
+const GEMINI_PATTERNS = {
+    context_overflow: [/token limit exceeded/i, /maximum input size/i, /context too long/i],
+    rate_limit: [/rate limit exceeded/i, /quota exceeded/i, /too many requests/i],
+    quota_exceeded: [/quota exceeded/i, /billing limit/i, /project quota/i],
+};
+// Generic patterns
+const GENERIC_PATTERNS = {
+    context_overflow: [
+        /context.*overflow/i,
+        /context.*exceeded/i,
+        /token.*limit/i,
+        /maximum.*context/i,
+        /message too long/i,
+        /input too long/i,
+    ],
+    rate_limit: [/rate.?limit/i, /too many requests/i, /throttled/i, /429/i],
+    timeout: [/timeout/i, /timed out/i, /etimedout/i],
+    network_error: [/network error/i, /econnreset/i, /econnrefused/i, /enotfound/i, /network/i],
+    server_error: [/server error/i, /internal error/i, /500/i, /502/i, /503/i, /504/i],
+    compaction_failure: [/compaction failed/i, /summarization failed/i, /failed to compact/i],
+};
+/**
+ * Classify an error based on message patterns
+ */
+export function classifyError(error) {
+    if (!(error instanceof Error)) {
+        return {
+            type: "unknown",
+            retryable: false,
+            message: String(error),
+        };
+    }
+    const message = error.message.toLowerCase();
+    // Check context overflow first (highest priority for LLM errors)
+    for (const pattern of [
+        ...OPENAI_PATTERNS.context_overflow,
+        ...ANTHROPIC_PATTERNS.context_overflow,
+        ...GEMINI_PATTERNS.context_overflow,
+        ...GENERIC_PATTERNS.context_overflow,
+    ]) {
+        if (pattern.test(message)) {
+            return {
+                type: "context_overflow",
+                retryable: false,
+                message: error.message,
+            };
+        }
+    }
+    // Check quota exceeded (before rate limit to avoid false positives)
+    for (const pattern of GEMINI_PATTERNS.quota_exceeded) {
+        if (pattern.test(message)) {
+            return {
+                type: "quota_exceeded",
+                retryable: false,
+                message: error.message,
+            };
+        }
+    }
+    // Check rate limit
+    for (const pattern of [
+        ...OPENAI_PATTERNS.rate_limit,
+        ...ANTHROPIC_PATTERNS.rate_limit,
+        ...GEMINI_PATTERNS.rate_limit,
+        ...GENERIC_PATTERNS.rate_limit,
+    ]) {
+        if (pattern.test(message)) {
+            return {
+                type: "rate_limit",
+                retryable: true,
+                message: error.message,
+            };
+        }
+    }
+    // Check authentication
+    for (const pattern of [...OPENAI_PATTERNS.authentication, ...ANTHROPIC_PATTERNS.authentication]) {
+        if (pattern.test(message)) {
+            return {
+                type: "authentication",
+                retryable: false,
+                message: error.message,
+            };
+        }
+    }
+    // Check authorization (403-like errors)
+    for (const pattern of ANTHROPIC_PATTERNS.authorization) {
+        if (pattern.test(message)) {
+            return {
+                type: "authorization",
+                retryable: false,
+                message: error.message,
+            };
+        }
+    }
+    // Check timeout
+    for (const pattern of GENERIC_PATTERNS.timeout) {
+        if (pattern.test(message)) {
+            return {
+                type: "timeout",
+                retryable: true,
+                message: error.message,
+            };
+        }
+    }
+    // Check network errors
+    for (const pattern of GENERIC_PATTERNS.network_error) {
+        if (pattern.test(message)) {
+            return {
+                type: "network_error",
+                retryable: true,
+                message: error.message,
+            };
+        }
+    }
+    // Check server errors
+    for (const pattern of GENERIC_PATTERNS.server_error) {
+        if (pattern.test(message)) {
+            return {
+                type: "server_error",
+                retryable: true,
+                message: error.message,
+            };
+        }
+    }
+    // Check compaction failure
+    for (const pattern of GENERIC_PATTERNS.compaction_failure) {
+        if (pattern.test(message)) {
+            return {
+                type: "compaction_failure",
+                retryable: true,
+                message: error.message,
+            };
+        }
+    }
+    // Default to unknown
+    return {
+        type: "unknown",
+        retryable: false,
+        message: error.message,
+    };
+}
+/**
+ * Check if error is a context overflow error
+ */
+export function isContextOverflowError(error) {
+    const classification = classifyError(error);
+    return classification.type === "context_overflow";
+}
+/**
+ * Check if error is a rate limit error
+ */
+export function isRateLimitError(error) {
+    const classification = classifyError(error);
+    return classification.type === "rate_limit";
+}
+/**
+ * Check if error is a compaction failure
+ */
+export function isCompactionFailureError(error) {
+    const classification = classifyError(error);
+    return classification.type === "compaction_failure";
+}
+/**
+ * Check if error is likely a context overflow (heuristic)
+ */
+export function isLikelyContextOverflowError(error) {
+    if (!(error instanceof Error)) {
+        return false;
+    }
+    const message = error.message.toLowerCase();
+    // Check for common indicators
+    const indicators = [
+        "context",
+        "token",
+        "length",
+        "exceeded",
+        "maximum",
+        "limit",
+        "too long",
+        "too many",
+    ];
+    const score = indicators.reduce((acc, indicator) => {
+        return acc + (message.includes(indicator) ? 1 : 0);
+    }, 0);
+    // If 3+ indicators present, likely context overflow
+    return score >= 3;
+}
+/**
+ * Get retry delay for error type
+ */
+export function getRetryDelayForError(error, attempt) {
+    const classification = classifyError(error);
+    switch (classification.type) {
+        case "rate_limit":
+            // Rate limits: start with 2s, double each attempt
+            return Math.min(2000 * 2 ** attempt, 60000);
+        case "timeout":
+        case "network_error":
+            // Network issues: start with 500ms
+            return Math.min(500 * 2 ** attempt, 30000);
+        case "server_error":
+            // Server errors: start with 1s
+            return Math.min(1000 * 2 ** attempt, 30000);
+        case "compaction_failure":
+            // Compaction: shorter delays
+            return Math.min(500 * 2 ** attempt, 5000);
+        default:
+            return 1000;
+    }
+}
+/**
+ * Format error for logging (sanitized)
+ */
+export function formatErrorForLogging(error) {
+    if (error instanceof Error) {
+        const classification = classifyError(error);
+        return `[${classification.type}] ${classification.message}`;
+    }
+    return `[unknown] ${String(error)}`;
+}