agent-relay 1.1.0 → 1.2.3

package/dist/cloud/services/scaling-policy.js

@@ -0,0 +1,415 @@
+/**
+ * Scaling Policy Service
+ *
+ * Defines rules and policies for auto-scaling workspaces based on:
+ * - Memory pressure
+ * - Agent count
+ * - CPU usage
+ * - Trend analysis
+ *
+ * Policies are configurable per user/plan tier.
+ */
+import { EventEmitter } from 'events';
+// Default thresholds by plan
+const DEFAULT_THRESHOLDS = {
+    free: {
+        memoryWarningBytes: 256 * 1024 * 1024, // 256MB
+        memoryCriticalBytes: 512 * 1024 * 1024, // 512MB
+        memoryScaleUpBytes: 400 * 1024 * 1024, // 400MB (no auto-scale for free)
+        memoryGrowthRateWarning: 5 * 1024 * 1024, // 5MB/min
+        memoryGrowthRateScaleUp: 10 * 1024 * 1024, // 10MB/min
+        agentsPerWorkspaceWarning: 3,
+        agentsPerWorkspaceMax: 5,
+        cpuWarningPercent: 70,
+        cpuScaleUpPercent: 85,
+        evaluationWindowMs: 5 * 60 * 1000, // 5 minutes
+        cooldownMs: 30 * 60 * 1000, // 30 minutes (free tier)
+    },
+    pro: {
+        memoryWarningBytes: 512 * 1024 * 1024, // 512MB
+        memoryCriticalBytes: 1024 * 1024 * 1024, // 1GB
+        memoryScaleUpBytes: 768 * 1024 * 1024, // 768MB
+        memoryGrowthRateWarning: 10 * 1024 * 1024, // 10MB/min
+        memoryGrowthRateScaleUp: 20 * 1024 * 1024, // 20MB/min
+        agentsPerWorkspaceWarning: 8,
+        agentsPerWorkspaceMax: 15,
+        cpuWarningPercent: 75,
+        cpuScaleUpPercent: 90,
+        evaluationWindowMs: 3 * 60 * 1000, // 3 minutes
+        cooldownMs: 10 * 60 * 1000, // 10 minutes
+    },
+    team: {
+        memoryWarningBytes: 768 * 1024 * 1024, // 768MB
+        memoryCriticalBytes: 1.5 * 1024 * 1024 * 1024, // 1.5GB
+        memoryScaleUpBytes: 1024 * 1024 * 1024, // 1GB
+        memoryGrowthRateWarning: 15 * 1024 * 1024, // 15MB/min
+        memoryGrowthRateScaleUp: 30 * 1024 * 1024, // 30MB/min
+        agentsPerWorkspaceWarning: 15,
+        agentsPerWorkspaceMax: 25,
+        cpuWarningPercent: 80,
+        cpuScaleUpPercent: 92,
+        evaluationWindowMs: 2 * 60 * 1000, // 2 minutes
+        cooldownMs: 5 * 60 * 1000, // 5 minutes
+    },
+    enterprise: {
+        memoryWarningBytes: 1024 * 1024 * 1024, // 1GB
+        memoryCriticalBytes: 2 * 1024 * 1024 * 1024, // 2GB
+        memoryScaleUpBytes: 1.5 * 1024 * 1024 * 1024, // 1.5GB
+        memoryGrowthRateWarning: 20 * 1024 * 1024, // 20MB/min
+        memoryGrowthRateScaleUp: 50 * 1024 * 1024, // 50MB/min
+        agentsPerWorkspaceWarning: 25,
+        agentsPerWorkspaceMax: 50,
+        cpuWarningPercent: 85,
+        cpuScaleUpPercent: 95,
+        evaluationWindowMs: 1 * 60 * 1000, // 1 minute
+        cooldownMs: 2 * 60 * 1000, // 2 minutes
+    },
+};
+// Default policies - ordered by priority (higher = evaluated first)
+// In-workspace scaling is preferred over adding new workspaces (more efficient)
+const DEFAULT_POLICIES = [
+    // === In-Workspace Scaling (Higher Priority) ===
+    {
+        id: 'agent-limit-increase',
+        name: 'Increase Agent Limit',
+        description: 'Increase max agents when approaching limit within single workspace',
+        enabled: true,
+        priority: 150, // Higher priority - try this before adding workspaces
+        conditions: [
+            { metric: 'agent_count', operator: 'gte', value: 0.85 }, // 85% of max agents
+            { metric: 'workspace_count', operator: 'eq', value: 1 }, // Only 1 workspace
+        ],
+        action: { type: 'increase_agent_limit', percentage: 50 }, // Increase limit by 50%
+        maxInstances: 10,
+        minInstances: 1,
+    },
+    {
+        id: 'workspace-resize-up',
+        name: 'Resize Workspace Up',
+        description: 'Vertically scale workspace when memory is high',
+        enabled: true,
+        priority: 140, // Higher priority than horizontal scaling
+        conditions: [
+            { metric: 'memory_usage', operator: 'gte', value: 0.75, duration: 120000 }, // 75% for 2min
+            { metric: 'workspace_count', operator: 'eq', value: 1 }, // Only 1 workspace
+        ],
+        action: { type: 'resize_up', percentage: 100 }, // Double resources
+        maxInstances: 10,
+        minInstances: 1,
+    },
+    {
+        id: 'cpu-pressure-resize',
+        name: 'CPU Pressure Resize',
+        description: 'Resize workspace when CPU is consistently high',
+        enabled: true,
+        priority: 135,
+        conditions: [
+            { metric: 'cpu_usage', operator: 'gte', value: 0.85, duration: 180000 }, // 85% for 3min
+        ],
+        action: { type: 'resize_up', percentage: 50 }, // 50% more resources
+        maxInstances: 10,
+        minInstances: 1,
+    },
+    {
+        id: 'workspace-resize-down',
+        name: 'Resize Workspace Down',
+        description: 'Reduce workspace resources when underutilized',
+        enabled: true,
+        priority: 45, // Lower priority
+        conditions: [
+            { metric: 'memory_usage', operator: 'lt', value: 0.15, duration: 900000 }, // Under 15% for 15min
+            { metric: 'cpu_usage', operator: 'lt', value: 0.1, duration: 900000 }, // Under 10% CPU
+        ],
+        action: { type: 'resize_down', percentage: 50 }, // Halve resources
+        maxInstances: 10,
+        minInstances: 1,
+    },
+    // === Horizontal Scaling (Add/Remove Workspaces) ===
+    {
+        id: 'memory-pressure-scale-up',
+        name: 'Memory Pressure Scale Up',
+        description: 'Add workspace when memory exceeds threshold across all workspaces',
+        enabled: true,
+        priority: 100,
+        conditions: [
+            { metric: 'memory_usage', operator: 'gte', value: 0.8, duration: 60000 }, // 80% for 1min
+        ],
+        action: { type: 'scale_up', targetCount: 1 },
+        maxInstances: 10,
+        minInstances: 1,
+    },
+    {
+        id: 'memory-trend-scale-up',
+        name: 'Memory Trend Scale Up',
+        description: 'Add workspace when memory growth rate is high',
+        enabled: true,
+        priority: 90,
+        conditions: [
+            { metric: 'memory_trend', operator: 'gte', value: 1.0, duration: 180000 }, // At threshold for 3min
+        ],
+        action: { type: 'scale_up', targetCount: 1 },
+        maxInstances: 10,
+        minInstances: 1,
+    },
+    {
+        id: 'agent-count-scale-up',
+        name: 'Agent Count Scale Up',
+        description: 'Add workspace when agent count is high across all workspaces',
+        enabled: true,
+        priority: 80,
+        conditions: [
+            { metric: 'agent_count', operator: 'gte', value: 0.9 }, // 90% of max agents
+            { metric: 'workspace_count', operator: 'gte', value: 1 }, // Already tried in-workspace scaling
+        ],
+        action: { type: 'scale_up', targetCount: 1 },
+        maxInstances: 10,
+        minInstances: 1,
+    },
+    // === Rebalancing ===
+    {
+        id: 'agent-rebalance',
+        name: 'Agent Rebalance',
+        description: 'Redistribute agents when load is uneven across workspaces',
+        enabled: true,
+        priority: 60,
+        conditions: [
+            { metric: 'workspace_count', operator: 'gte', value: 2 }, // Multiple workspaces
+        ],
+        action: { type: 'rebalance' },
+        maxInstances: 10,
+        minInstances: 1,
+    },
+    // === Scale Down ===
+    {
+        id: 'low-usage-scale-down',
+        name: 'Low Usage Scale Down',
+        description: 'Remove workspace when usage is low',
+        enabled: true,
+        priority: 50,
+        conditions: [
+            { metric: 'memory_usage', operator: 'lt', value: 0.2, duration: 600000 }, // Under 20% for 10min
+            { metric: 'workspace_count', operator: 'gt', value: 1 }, // More than 1 workspace
+        ],
+        action: { type: 'scale_down', targetCount: 1 },
+        maxInstances: 10,
+        minInstances: 1,
+    },
+];
+export class ScalingPolicyService extends EventEmitter {
+    thresholds = new Map();
+    policies = new Map();
+    conditionHistory = new Map();
+    constructor() {
+        super();
+        // Initialize with defaults
+        for (const [plan, thresholds] of Object.entries(DEFAULT_THRESHOLDS)) {
+            this.thresholds.set(plan, thresholds);
+        }
+    }
+    /**
+     * Get thresholds for a plan tier
+     */
+    getThresholds(plan) {
+        return this.thresholds.get(plan) || this.thresholds.get('free');
+    }
+    /**
+     * Set custom thresholds for a plan
+     */
+    setThresholds(plan, thresholds) {
+        const current = this.getThresholds(plan);
+        this.thresholds.set(plan, { ...current, ...thresholds });
+    }
+    /**
+     * Get policies for a user (default + custom)
+     */
+    getPolicies(userId) {
+        const userPolicies = this.policies.get(userId) || [];
+        return [...DEFAULT_POLICIES, ...userPolicies].sort((a, b) => b.priority - a.priority);
+    }
+    /**
+     * Add custom policy for a user
+     */
+    addPolicy(userId, policy) {
+        const existing = this.policies.get(userId) || [];
+        existing.push(policy);
+        this.policies.set(userId, existing);
+    }
+    /**
+     * Evaluate scaling decision based on current context
+     */
+    evaluate(context) {
+        const thresholds = this.getThresholds(context.plan);
+        const policies = this.getPolicies(context.userId);
+        // Calculate aggregate metrics
+        const metrics = this.calculateAggregateMetrics(context, thresholds);
+        // Check cooldown
+        if (context.lastScalingAction) {
+            const timeSinceLastScale = Date.now() - context.lastScalingAction.getTime();
+            if (timeSinceLastScale < thresholds.cooldownMs) {
+                return {
+                    shouldScale: false,
+                    action: null,
+                    reason: `Cooldown active (${Math.round((thresholds.cooldownMs - timeSinceLastScale) / 1000)}s remaining)`,
+                    triggeredPolicy: null,
+                    metrics,
+                    timestamp: new Date(),
+                };
+            }
+        }
+        // Evaluate policies in priority order
+        for (const policy of policies) {
+            if (!policy.enabled)
+                continue;
+            const conditionsMet = this.evaluateConditions(policy.conditions, metrics, thresholds, context.userId);
+            if (conditionsMet) {
+                // Check instance limits for horizontal scaling only
+                if (policy.action.type === 'scale_up') {
+                    // Block if at workspace limit (for adding new workspaces)
+                    if (context.currentWorkspaceCount >= context.maxWorkspaces) {
+                        continue; // Try next policy (could be in-workspace scaling)
+                    }
+                    if (context.currentWorkspaceCount >= policy.maxInstances) {
+                        continue;
+                    }
+                }
+                if (policy.action.type === 'scale_down' && context.currentWorkspaceCount <= policy.minInstances) {
+                    continue;
+                }
+                this.emit('scaling_decision', {
+                    userId: context.userId,
+                    policy: policy.id,
+                    action: policy.action,
+                    metrics,
+                });
+                return {
+                    shouldScale: true,
+                    action: policy.action,
+                    reason: policy.description,
+                    triggeredPolicy: policy.id,
+                    metrics,
+                    timestamp: new Date(),
+                };
+            }
+        }
+        return {
+            shouldScale: false,
+            action: null,
+            reason: 'No scaling conditions met',
+            triggeredPolicy: null,
+            metrics,
+            timestamp: new Date(),
+        };
+    }
+    /**
+     * Calculate aggregate metrics from workspace metrics
+     */
+    calculateAggregateMetrics(context, thresholds) {
+        const workspaces = context.workspaceMetrics;
+        if (workspaces.length === 0) {
+            return {
+                memory_usage: 0,
+                memory_trend: 0,
+                agent_count: 0,
+                cpu_usage: 0,
+                workspace_count: 0,
+                total_memory_bytes: 0,
+                total_agents: 0,
+            };
+        }
+        const totalMemory = workspaces.reduce((sum, w) => sum + w.totalMemoryBytes, 0);
+        const avgTrend = workspaces.reduce((sum, w) => sum + w.memoryTrendPerMinute, 0) / workspaces.length;
+        const totalAgents = workspaces.reduce((sum, w) => sum + w.agentCount, 0);
+        const avgCpu = workspaces.reduce((sum, w) => sum + w.cpuPercent, 0) / workspaces.length;
+        // Normalized metrics (0-1 scale relative to thresholds)
+        return {
+            memory_usage: totalMemory / (thresholds.memoryScaleUpBytes * workspaces.length),
+            memory_trend: avgTrend / thresholds.memoryGrowthRateScaleUp,
+            agent_count: totalAgents / (thresholds.agentsPerWorkspaceMax * workspaces.length),
+            cpu_usage: avgCpu / thresholds.cpuScaleUpPercent,
+            workspace_count: context.currentWorkspaceCount,
+            total_memory_bytes: totalMemory,
+            total_agents: totalAgents,
+        };
+    }
+    /**
+     * Evaluate conditions with duration support
+     */
+    evaluateConditions(conditions, metrics, thresholds, userId) {
+        for (const condition of conditions) {
+            const metricValue = metrics[condition.metric];
+            if (metricValue === undefined)
+                continue;
+            const conditionMet = this.compareValues(metricValue, condition.operator, condition.value);
+            if (condition.duration) {
+                // Track condition history for duration-based evaluation
+                const historyKey = `${userId}:${condition.metric}`;
+                const history = this.conditionHistory.get(historyKey) || [];
+                // Add current value
+                history.push({ timestamp: new Date(), value: metricValue });
+                // Clean old entries
+                const cutoff = Date.now() - condition.duration;
+                const recentHistory = history.filter((h) => h.timestamp.getTime() > cutoff);
+                this.conditionHistory.set(historyKey, recentHistory);
+                // Check if condition has been met for the full duration
+                if (recentHistory.length === 0)
+                    return false;
+                const allMet = recentHistory.every((h) => this.compareValues(h.value, condition.operator, condition.value));
+                // Also check if we have enough history
+                const oldestEntry = recentHistory[0].timestamp.getTime();
+                const hasEnoughHistory = Date.now() - oldestEntry >= condition.duration * 0.8; // 80% of duration
+                if (!allMet || !hasEnoughHistory)
+                    return false;
+            }
+            else {
+                if (!conditionMet)
+                    return false;
+            }
+        }
+        return true;
+    }
+    /**
+     * Compare values based on operator
+     */
+    compareValues(actual, operator, target) {
+        switch (operator) {
+            case 'gt':
+                return actual > target;
+            case 'gte':
+                return actual >= target;
+            case 'lt':
+                return actual < target;
+            case 'lte':
+                return actual <= target;
+            case 'eq':
+                return actual === target;
+            default:
+                return false;
+        }
+    }
+    /**
+     * Get max workspaces for a plan
+     */
+    getMaxWorkspaces(plan) {
+        switch (plan) {
+            case 'free':
+                return 1;
+            case 'pro':
+                return 3;
+            case 'team':
+                return 10;
+            case 'enterprise':
+                return 50;
+            default:
+                return 1;
+        }
+    }
+}
+// Singleton instance
+let _scalingPolicyService = null;
+export function getScalingPolicyService() {
+    if (!_scalingPolicyService) {
+        _scalingPolicyService = new ScalingPolicyService();
+    }
+    return _scalingPolicyService;
+}
+//# sourceMappingURL=scaling-policy.js.map

package/dist/cloud/vault/index.js

@@ -3,7 +3,7 @@
  *
  * Secure storage for OAuth tokens with AES-256-GCM encryption.
  */
-import crypto from 'crypto';
+import * as crypto from 'crypto';
 import { getConfig } from '../config.js';
 import { db } from '../db/index.js';
 const ALGORITHM = 'aes-256-gcm';

package/dist/cloud/webhooks/index.d.ts

@@ -0,0 +1,24 @@
+/**
+ * Generic Webhook System
+ *
+ * A configurable webhook system that can handle events from any source
+ * (GitHub, GitLab, Linear, Slack, Jira, etc.) and route them to agents.
+ *
+ * Components:
+ * - Parsers: Transform source-specific payloads to normalized events
+ * - Responders: Send responses back to source systems
+ * - Rules Engine: Match events to actions based on configuration
+ * - Router: Orchestrates the full webhook processing pipeline
+ */
+export * from './types.js';
+export { getParser, registerParser, parsers } from './parsers/index.js';
+export { githubParser } from './parsers/github.js';
+export { linearParser } from './parsers/linear.js';
+export { slackParser } from './parsers/slack.js';
+export { getResponder, registerResponder, responders } from './responders/index.js';
+export { githubResponder } from './responders/github.js';
+export { linearResponder } from './responders/linear.js';
+export { slackResponder, formatSlackBlocks } from './responders/slack.js';
+export { matchesRule, findMatchingRules, resolveActionTemplate, defaultRules, } from './rules-engine.js';
+export { processWebhook, getWebhookConfig, defaultSources, } from './router.js';
+//# sourceMappingURL=index.d.ts.map

package/dist/cloud/webhooks/index.js

@@ -0,0 +1,29 @@
+/**
+ * Generic Webhook System
+ *
+ * A configurable webhook system that can handle events from any source
+ * (GitHub, GitLab, Linear, Slack, Jira, etc.) and route them to agents.
+ *
+ * Components:
+ * - Parsers: Transform source-specific payloads to normalized events
+ * - Responders: Send responses back to source systems
+ * - Rules Engine: Match events to actions based on configuration
+ * - Router: Orchestrates the full webhook processing pipeline
+ */
+// Types
+export * from './types.js';
+// Parsers
+export { getParser, registerParser, parsers } from './parsers/index.js';
+export { githubParser } from './parsers/github.js';
+export { linearParser } from './parsers/linear.js';
+export { slackParser } from './parsers/slack.js';
+// Responders
+export { getResponder, registerResponder, responders } from './responders/index.js';
+export { githubResponder } from './responders/github.js';
+export { linearResponder } from './responders/linear.js';
+export { slackResponder, formatSlackBlocks } from './responders/slack.js';
+// Rules Engine
+export { matchesRule, findMatchingRules, resolveActionTemplate, defaultRules, } from './rules-engine.js';
+// Router
+export { processWebhook, getWebhookConfig, defaultSources, } from './router.js';
+//# sourceMappingURL=index.js.map

package/dist/cloud/webhooks/parsers/github.d.ts

@@ -0,0 +1,8 @@
+/**
+ * GitHub Webhook Parser
+ *
+ * Transforms GitHub webhook payloads into normalized events.
+ */
+import type { WebhookParser } from '../types.js';
+export declare const githubParser: WebhookParser;
+//# sourceMappingURL=github.d.ts.map