Spaces:
Sleeping
Sleeping
File size: 3,656 Bytes
b6ecafa | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 | import { NextRequest, NextResponse } from 'next/server'
import { requireRole } from '@/lib/auth'
import { readLimiter } from '@/lib/rate-limit'
import { logger } from '@/lib/logger'
import {
analyzeTokenEfficiency,
analyzeToolPatterns,
getFleetBenchmarks,
generateRecommendations,
} from '@/lib/agent-optimizer'
export async function GET(request: NextRequest) {
const auth = requireRole(request, 'operator')
if ('error' in auth) return NextResponse.json({ error: auth.error }, { status: auth.status })
const rateCheck = readLimiter(request)
if (rateCheck) return rateCheck
try {
const { searchParams } = new URL(request.url)
const agent = searchParams.get('agent')
const hours = parseInt(searchParams.get('hours') || '24', 10)
const workspaceId = auth.user.workspace_id ?? 1
if (!agent) {
return NextResponse.json({ error: 'Missing required parameter: agent' }, { status: 400 })
}
const efficiency = analyzeTokenEfficiency(agent, hours, workspaceId)
const toolPatterns = analyzeToolPatterns(agent, hours, workspaceId)
const fleet = getFleetBenchmarks(workspaceId)
const recommendations = generateRecommendations(agent, workspaceId)
// Calculate fleet percentile for tokens per session
const fleetTokens = fleet
.map(f => f.tokensPerTask)
.filter(t => t > 0)
.sort((a, b) => a - b)
const agentTokensPerTask = efficiency.sessionsCount > 0 ? efficiency.avgTokensPerSession : 0
const percentile = fleetTokens.length > 0
? Math.round((fleetTokens.filter(t => t >= agentTokensPerTask).length / fleetTokens.length) * 100)
: 50
// Fleet average cost
const fleetAvgCost = fleet.length > 0
? fleet.reduce((sum, f) => sum + f.costPerTask, 0) / fleet.length
: 0
// Tool analysis
const mostUsed = toolPatterns.topTools.slice(0, 5)
const leastEffective = toolPatterns.topTools
.filter(t => t.successRate < 80)
.sort((a, b) => a.successRate - b.successRate)
.slice(0, 5)
// Performance from fleet benchmarks
const agentBenchmark = fleet.find(f => f.agentName === agent)
return NextResponse.json({
agent,
analyzedAt: new Date().toISOString(),
efficiency: {
tokensPerTask: agentTokensPerTask,
fleetAverage: fleetTokens.length > 0
? Math.round(fleetTokens.reduce((a, b) => a + b, 0) / fleetTokens.length)
: 0,
percentile,
trend: efficiency.totalTokens,
costPerTask: efficiency.avgCostPerSession,
},
toolPatterns: {
mostUsed: mostUsed.map(t => ({
name: t.toolName,
count: t.count,
successRate: t.successRate,
})),
leastEffective: leastEffective.map(t => ({
name: t.toolName,
count: t.count,
successRate: t.successRate,
})),
unusedCapabilities: [],
},
performance: {
taskCompletionRate: agentBenchmark?.tasksCompleted ?? 0,
avgTaskDuration: toolPatterns.avgDurationMs,
errorRate: toolPatterns.failureRate,
fleetRanking: fleet.findIndex(f => f.agentName === agent) + 1 || fleet.length + 1,
},
recommendations: recommendations.map(r => ({
category: r.category,
priority: r.severity,
title: r.category.charAt(0).toUpperCase() + r.category.slice(1) + ' issue',
description: r.message,
expectedImpact: r.metric ?? null,
})),
})
} catch (error) {
logger.error({ err: error }, 'GET /api/agents/optimize error')
return NextResponse.json({ error: 'Internal server error' }, { status: 500 })
}
}
|