langgraph-api 0.12.0.dev11__tar.gz → 0.12.0.dev12__tar.gz

{langgraph_api-0.12.0.dev11 → langgraph_api-0.12.0.dev12}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: langgraph-api
-Version: 0.12.0.dev11
+Version: 0.12.0.dev12
 Author-email: Will Fu-Hinthorn <will@langchain.dev>, Josh Rogers <josh@langchain.dev>, Parker Rule <parker@langchain.dev>
 License: Elastic-2.0
 License-File: LICENSE

{langgraph_api-0.12.0.dev11 → langgraph_api-0.12.0.dev12}/benchmark/.gitignore

@@ -9,5 +9,3 @@ staircase_*.json
 
 # Node.js dependencies
 node_modules/
-capacity_summary.json
-cap-bench-results/

langgraph_api-0.12.0.dev12/benchmark/Makefile

@@ -0,0 +1,30 @@
+# Benchmark commands
+
+benchmark-ramp:
+	make benchmark-reset
+	npm install
+	npm run build
+	k6 run --out json=raw_data_$(shell date +%Y-%m-%dT%H-%M-%S).json --system-tags=[] ramp.js
+
+benchmark-charts:
+	npm install
+	node graphs.js $(shell ls -t raw_data_*.json | head -1) true
+
+benchmark-reset:
+	node clean-cli.js
+
+benchmark-new-revision:
+	node update-revision.js
+
+benchmark-staircase:
+	python staircase.py
+
+benchmark-staircase-reset:
+	python ../../scripts/reset_deployment.py -a $(BENCHMARK_APP_NAME) -n $(BENCHMARK_NAMESPACE) -db $(BENCHMARK_DB_INSTANCE)
+	python staircase.py
+
+benchmark-clean:
+	rm -f results_*.json summary_*.json raw_data_*.json staircase_*.json *_chart_*.png
+
+benchmark-clean-charts:
+	rm -f *_chart_*.png

{langgraph_api-0.12.0.dev11 → langgraph_api-0.12.0.dev12}/benchmark/README.md

@@ -20,23 +20,6 @@ This directory contains k6 + Python/Node benchmark tooling for LangGraph API.
   - **Example:**
     - `BENCHMARK_PROFILE=parallel-small START_LOAD=30 STEP_SIZE=15 NUM_STEPS=10 make benchmark-staircase`
 
-- `make benchmark-capacity` - Runs binary-search capacity tests and writes `capacity_summary.json`.
-  - **Environment variables:**
-    - `CLUSTER_NAME`
-    - `WORKLOAD_NAMES` (comma-separated) or `WORKLOAD_NAME`
-  - **Support note:** `BENCHMARK_TYPE` is supported but usually determined by profile/workload defaults.
-  - **Example:**
-    - `CLUSTER_NAME=dr-1-node WORKLOAD_NAMES=parallel-small,parallel-medium make benchmark-capacity`
-
-- `make benchmark-mixed` - Runs quick and long workloads concurrently to measure contention/fairness.
-  - **Environment variables:**
-    - `SLOWEST_N`
-    - Quick workload knobs: `QUICK_VUS`, `QUICK_ITERATIONS`, `QUICK_STEPS`, `QUICK_DATA_SIZE`, `QUICK_MAX_WAIT_SECONDS`, `QUICK_POLL_INTERVAL`
-    - Long workload knobs: `LONG_VUS`, `LONG_ITERATIONS`, `LONG_STEPS`, `LONG_DELAY`, `LONG_DATA_SIZE`, `LONG_MAX_WAIT_SECONDS`, `LONG_POLL_INTERVAL`
-  - **Support note:** Does **not** support benchmark universal/profile/context knobs (`BENCHMARK_TYPE`, `BENCHMARK_PROFILE`, `RUN_MODE`, `STREAM_RESUMABLE`, or context overrides). Only mixed-workload vars + generic `BASE_URL`/`LANGSMITH_API_KEY` apply.
-  - **Example:**
-    - `BASE_URL=http://localhost:9123 QUICK_VUS=300 LONG_VUS=8 LONG_STEPS=60 make benchmark-mixed`
-
 - `make benchmark-charts` - Builds charts from the latest `raw_data_*.json` ramp output. Only works for ramp.
 - `make benchmark-clean` - Deletes local benchmark result/chart artifacts.
 - `make benchmark-reset` - Cleans benchmark server state via `clean-cli.js`.
@@ -70,7 +53,7 @@ Profiles are quick ways to configure the benchmark graph for any run depending o
 Each profile defines:
 
 - `context` (delay/steps/checkpoint/streaming/burst knobs)
-- `capacity` defaults (`runMode`, `rampEndBase`, `runExecutionTimeoutSeconds`)
+- `runMode` default (`stateless` or `stateful`, can be overridden by `RUN_MODE`)
 - `resumable` default (can be overridden by `STREAM_RESUMABLE`)
 
 ## Core environment knobs
@@ -110,15 +93,6 @@ BENCHMARK_PROFILE=streaming-long BENCHMARK_TYPE=stream_write make benchmark-ramp
 # Staircase benchmark
 make benchmark-staircase
 
-# Capacity benchmark (single workload)
-CLUSTER_NAME=dr-1-node WORKLOAD_NAME=parallel-small make benchmark-capacity
-
-# Capacity benchmark (multiple workloads)
-CLUSTER_NAME=dr-1-node WORKLOAD_NAMES=parallel-small,parallel-medium make benchmark-capacity
-
-# Mixed workload benchmark
-make benchmark-mixed
-
 # Build charts from latest ramp raw_data file
 make benchmark-charts
 
@@ -135,16 +109,10 @@ make benchmark-clean
   - `*_chart_*.png` (when chart generation is run)
 - Staircase:
   - `staircase_summary.json`
-- Capacity:
-  - `capacity_summary.json`
-- Mixed:
-  - `mixed_workload_*.json`
 
 ## CI workflows
 
 - Daily ramp benchmark: `.github/workflows/benchmark.yml`
-- Capacity benchmark matrix: `.github/workflows/benchmark-capacity.yml`
-- Mixed benchmark matrix: `.github/workflows/benchmark-mixed.yml`
 
 ## References
 

{langgraph_api-0.12.0.dev11 → langgraph_api-0.12.0.dev12}/benchmark/benchmark-runners/benchmark_profiles.ts

@@ -32,16 +32,10 @@ export interface BenchmarkProfileEnv {
   BURST_SIZE?: string;
 }
 
-export interface CapacityProfileSettings {
-  runMode: 'stateless' | 'stateful';
-  rampEndBase: number;
-  runExecutionTimeoutSeconds: number;
-}
-
 export interface BenchmarkProfile {
   context: BenchmarkContext;
   resumable: boolean;
-  capacity: CapacityProfileSettings;
+  runMode: 'stateless' | 'stateful';
 }
 
 export interface ParsedBenchmarkProfile {
@@ -50,7 +44,6 @@ export interface ParsedBenchmarkProfile {
   runMode: 'stateless' | 'stateful';
   resumable: boolean;
   context: BenchmarkContext;
-  capacity: CapacityProfileSettings;
 }
 
 const DEFAULT_PROFILE_NAME = 'default';
@@ -58,6 +51,7 @@ const DEFAULT_PROFILE_NAME = 'default';
 export const BENCHMARK_PROFILES: Record<string, BenchmarkProfile> = {
   default: {
     resumable: false,
+    runMode: 'stateless',
     context: {
       delay: 0,
       delay_jitter_ratio: 0.2,
@@ -71,14 +65,10 @@ export const BENCHMARK_PROFILES: Record<string, BenchmarkProfile> = {
       burst_probability: 0,
       burst_size: 0,
     },
-    capacity: {
-      runMode: 'stateless',
-      rampEndBase: 200,
-      runExecutionTimeoutSeconds: 60,
-    },
   },
   'streaming-long': {
     resumable: true,
+    runMode: 'stateful',
     context: {
       delay: 0.5,
       delay_jitter_ratio: 0.2,
@@ -92,14 +82,10 @@ export const BENCHMARK_PROFILES: Record<string, BenchmarkProfile> = {
       burst_probability: 0.05,
       burst_size: 2 * 1024 * 1024,
     },
-    capacity: {
-      runMode: 'stateful',
-      rampEndBase: 20,
-      runExecutionTimeoutSeconds: 1000,
-    },
   },
   'stateless-parallel-small': {
     resumable: false,
+    runMode: 'stateless',
     context: {
       delay: 0,
       delay_jitter_ratio: 0.2,
@@ -113,14 +99,10 @@ export const BENCHMARK_PROFILES: Record<string, BenchmarkProfile> = {
       burst_probability: 0,
       burst_size: 0,
     },
-    capacity: {
-      runMode: 'stateless',
-      rampEndBase: 200,
-      runExecutionTimeoutSeconds: 60,
-    },
   },
   'stateless-parallel-medium': {
     resumable: false,
+    runMode: 'stateless',
     context: {
       delay: 0,
       delay_jitter_ratio: 0.2,
@@ -134,14 +116,10 @@ export const BENCHMARK_PROFILES: Record<string, BenchmarkProfile> = {
       burst_probability: 0,
       burst_size: 0,
     },
-    capacity: {
-      runMode: 'stateless',
-      rampEndBase: 200,
-      runExecutionTimeoutSeconds: 60,
-    },
   },
   'parallel-small': {
     resumable: false,
+    runMode: 'stateful',
     context: {
       delay: 0,
       delay_jitter_ratio: 0.2,
@@ -155,14 +133,10 @@ export const BENCHMARK_PROFILES: Record<string, BenchmarkProfile> = {
       burst_probability: 0,
       burst_size: 0,
     },
-    capacity: {
-      runMode: 'stateful',
-      rampEndBase: 200,
-      runExecutionTimeoutSeconds: 60,
-    },
   },
   'parallel-medium': {
     resumable: false,
+    runMode: 'stateful',
     context: {
       delay: 0,
       delay_jitter_ratio: 0.2,
@@ -176,11 +150,6 @@ export const BENCHMARK_PROFILES: Record<string, BenchmarkProfile> = {
       burst_probability: 0,
       burst_size: 0,
     },
-    capacity: {
-      runMode: 'stateful',
-      rampEndBase: 200,
-      runExecutionTimeoutSeconds: 60,
-    },
   },
 };
 
@@ -254,13 +223,10 @@ export function get_profile(env: BenchmarkProfileEnv): ParsedBenchmarkProfile {
   return {
     name,
     benchmarkType: env.BENCHMARK_TYPE || 'wait_write',
-    runMode: env.RUN_MODE ? (env.RUN_MODE === 'stateful' ? 'stateful' : 'stateless') : profile.capacity.runMode,
+    runMode: env.RUN_MODE ? (env.RUN_MODE === 'stateful' ? 'stateful' : 'stateless') : profile.runMode,
     resumable: env.STREAM_RESUMABLE != null && env.STREAM_RESUMABLE !== ''
       ? parseOptionalBool(env.STREAM_RESUMABLE) ?? profile.resumable
       : profile.resumable,
     context,
-    capacity: profile.capacity,
   };
 }
-
-

langgraph_api-0.12.0.dev12/langgraph_api/__init__.py

@@ -0,0 +1 @@
+__version__ = "0.12.0.dev12"

langgraph_api-0.12.0.dev11/benchmark/Makefile

@@ -1,93 +0,0 @@
-# Benchmark commands
-
-# Capacity benchmark
-CLUSTER_NAME ?= dr-1-node
-# WORKLOAD_NAMES supports comma-separated list (e.g., "parallel-small,parallel-tiny")
-# Falls back to WORKLOAD_NAME for backwards compatibility
-WORKLOAD_NAMES ?= $(WORKLOAD_NAME)
-WORKLOAD_NAME ?= parallel-small
-
-# For other benchmarks
-BASE_URL ?= https://benchmark-dr-s-2799835ad04b501a95044223ae72ced7.staging.langgraph.app
-SLOWEST_N ?= 2
-RAMP_MULTIPLIER ?= 2
-WAIT_SECONDS ?= 60
-CLEAR_DELAY_SECONDS ?= 5
-DATA_SIZE ?= 1000
-DELAY ?= 0
-EXPAND ?= 1
-STEPS ?= 100
-
-
-benchmark-ramp:
-	make benchmark-reset
-	npm install
-	npm run build
-	k6 run --out json=raw_data_$(shell date +%Y-%m-%dT%H-%M-%S).json --system-tags=[] ramp.js
-
-benchmark-capacity:
-	rm -f capacity_summary.json
-	npm install
-	npm run build
-	LANGSMITH_API_KEY=$(LANGSMITH_API_KEY) \
-	CLUSTER_NAME=$(CLUSTER_NAME) \
-	WORKLOAD_NAMES=$(WORKLOAD_NAMES) \
-	node capacity_runner.mjs
-
-benchmark-charts:
-	npm install
-	node graphs.js $(shell ls -t raw_data_*.json | head -1) true
-
-benchmark-reset:
-	node clean-cli.js
-
-benchmark-new-revision:
-	node update-revision.js
-
-benchmark-staircase:
-	python staircase.py
-
-benchmark-staircase-reset:
-	python ../../scripts/reset_deployment.py -a $(BENCHMARK_APP_NAME) -n $(BENCHMARK_NAMESPACE) -db $(BENCHMARK_DB_INSTANCE)
-	python staircase.py
-
-benchmark-clean:
-	rm -f results_*.json summary_*.json raw_data_*.json staircase_*.json *_chart_*.png
-
-benchmark-clean-charts:
-	rm -f *_chart_*.png
-
-# Mixed workload benchmark - tests quick runs competing with long runs
-QUICK_VUS ?= 250
-QUICK_ITERATIONS ?= 1000
-QUICK_STEPS ?= 5
-QUICK_DATA_SIZE ?= 100
-QUICK_MAX_WAIT_SECONDS ?= 120
-QUICK_POLL_INTERVAL ?= 2
-LONG_VUS ?= 5
-LONG_ITERATIONS ?= 10
-LONG_STEPS ?= 50
-LONG_DELAY ?= 1
-LONG_DATA_SIZE ?= 10000
-LONG_MAX_WAIT_SECONDS ?= 200
-LONG_POLL_INTERVAL ?= 5
-
-benchmark-mixed:
-	rm -f mixed_workload_*.json
-	npm install
-	BASE_URL=$(BASE_URL) \
-	SLOWEST_N=$(SLOWEST_N) \
-	QUICK_VUS=$(QUICK_VUS) \
-	QUICK_ITERATIONS=$(QUICK_ITERATIONS) \
-	QUICK_STEPS=$(QUICK_STEPS) \
-	QUICK_DATA_SIZE=$(QUICK_DATA_SIZE) \
-	QUICK_MAX_WAIT_SECONDS=$(QUICK_MAX_WAIT_SECONDS) \
-	QUICK_POLL_INTERVAL=$(QUICK_POLL_INTERVAL) \
-	LONG_VUS=$(LONG_VUS) \
-	LONG_ITERATIONS=$(LONG_ITERATIONS) \
-	LONG_STEPS=$(LONG_STEPS) \
-	LONG_DELAY=$(LONG_DELAY) \
-	LONG_DATA_SIZE=$(LONG_DATA_SIZE) \
-	LONG_MAX_WAIT_SECONDS=$(LONG_MAX_WAIT_SECONDS) \
-	LONG_POLL_INTERVAL=$(LONG_POLL_INTERVAL) \
-	node mixed_workload_runner.mjs

langgraph_api-0.12.0.dev11/benchmark/capacity_dd_report.py

@@ -1,96 +0,0 @@
-#!/usr/bin/env python3
-"""
-Send capacity benchmark results to Datadog.
-"""
-
-import json
-import os
-import sys
-
-from datadog_api_client import ApiClient, Configuration
-from datadog_api_client.v2.api.logs_api import LogsApi
-from datadog_api_client.v2.model.http_log import HTTPLog
-from datadog_api_client.v2.model.http_log_item import HTTPLogItem
-
-
-def send_capacity_results(summary_file: str):
-    """
-    Read capacity_summary.json and send to Datadog.
-
-    The summary file format is:
-    {
-        "clusterName": "dr-small",
-        "workloads": {
-            "parallel-small": {"maxSuccessfulTarget": 10, "avgExecutionLatencySeconds": 1.5, "p95ExecutionLatencySeconds": 2.0, "p99ExecutionLatencySeconds": 2.5},
-            "parallel-tiny": {"maxSuccessfulTarget": 20, "avgExecutionLatencySeconds": 0.8, "p95ExecutionLatencySeconds": 1.2, "p99ExecutionLatencySeconds": 1.5}
-        }
-    }
-
-    Args:
-        summary_file: Path to capacity_summary.json
-    """
-    # Get environment variables
-    dd_api_key = os.getenv("DD_API_KEY")
-    if not dd_api_key:
-        print("Error: DD_API_KEY not set", file=sys.stderr)  # noqa: T201
-        sys.exit(1)
-
-    dd_site = os.getenv("DD_SITE", "us5.datadoghq.com")
-
-    # Read summary file
-    try:
-        with open(summary_file) as f:
-            data = json.load(f)
-    except (OSError, json.JSONDecodeError) as e:
-        print(f"Error reading {summary_file}: {e}", file=sys.stderr)  # noqa: T201
-        sys.exit(1)
-
-    cluster_name = data.get("clusterName", "unknown")
-    workloads = data.get("workloads", {})
-
-    if not workloads:
-        print("Error: No workloads found in summary file", file=sys.stderr)  # noqa: T201
-        sys.exit(1)
-
-    # Configure Datadog client
-    configuration = Configuration()
-    configuration.server_variables["site"] = dd_site
-    configuration.api_key["apiKeyAuth"] = dd_api_key
-
-    # Create log items for each workload
-    log_items = []
-    for workload_name, result in workloads.items():
-        benchmark_data = {
-            "clusterName": cluster_name,
-            "workloadName": workload_name,
-            "maxSuccessfulTarget": result.get("maxSuccessfulTarget"),
-            "avgExecutionLatencySeconds": result.get("avgExecutionLatencySeconds"),
-            "p95ExecutionLatencySeconds": result.get("p95ExecutionLatencySeconds"),
-            "p99ExecutionLatencySeconds": result.get("p99ExecutionLatencySeconds"),
-            "benchmarkType": "capacity",
-        }
-
-        log_item = HTTPLogItem(
-            ddsource="capacity-benchmark",
-            ddtags=f"env:benchmarking,cluster:{cluster_name},workload:{workload_name}",
-            hostname=os.getenv("HOSTNAME", "github-actions"),
-            message=json.dumps(benchmark_data),
-            service="capacity-benchmark",
-        )
-        log_items.append(log_item)
-
-    # Send all logs to Datadog
-    with ApiClient(configuration) as api_client:
-        api_instance = LogsApi(api_client)
-        body = HTTPLog(log_items)
-        api_instance.submit_log(body=body)
-
-    print("Sent capacity results to Datadog")  # noqa: T201
-
-
-if __name__ == "__main__":
-    if len(sys.argv) < 2:
-        print("Usage: capacity_dd_report.py <capacity_summary.json>", file=sys.stderr)  # noqa: T201
-        sys.exit(1)
-
-    send_capacity_results(sys.argv[1])

langgraph_api-0.12.0.dev11/benchmark/capacity_k6.js

@@ -1,166 +0,0 @@
-import { check, fail } from "k6";
-import { Trend, Counter } from "k6/metrics";
-import { Benchmarks } from "./benchmark-runners/dist/benchmarks.js";
-import { get_profile } from "./benchmark-runners/dist/benchmark_profiles.js";
-
-// Metrics
-const runExecutionLatency = new Trend("run_execution_latency");
-// This object is taken in by runner.validate
-const errorMetrics = {
-  server_errors: new Counter("server_errors"),
-  timeout_errors: new Counter("timeout_errors"),
-  missing_message_errors: new Counter("missing_message_errors"),
-  api_errors: new Counter("api_errors"), // For __error__ responses with 200 status
-  other_errors: new Counter("other_errors"),
-};
-
-// Environment variables
-const BASE_URL = __ENV.BASE_URL;
-const LANGSMITH_API_KEY = __ENV.LANGSMITH_API_KEY;
-const TARGET = parseInt(__ENV.TARGET || "10");
-const RUN_EXECUTION_TIMEOUT_SECONDS = parseInt(
-  __ENV.RUN_EXECUTION_TIMEOUT_SECONDS || "10",
-);
-
-const profile = get_profile(__ENV);
-const BENCHMARK_TYPE = profile.benchmarkType;
-const BENCHMARK_PROFILE = profile.name;
-const RUN_MODE = profile.runMode;
-const EFFECTIVE_CONTEXT = profile.context;
-const RESUMABLE = profile.resumable;
-
-// K6 options
-export const options = {
-  // Include p99 in summary statistics (not included by default)
-  summaryTrendStats: ["avg", "min", "med", "max", "p(90)", "p(95)", "p(99)"],
-  scenarios: {
-    capacity_test: {
-      executor: "per-vu-iterations",
-      vus: TARGET,
-      iterations: 1, // Each VU executes once
-      maxDuration: `${RUN_EXECUTION_TIMEOUT_SECONDS + 30}s`,
-    },
-  },
-};
-
-// Build request params for benchmark runners
-function buildRequestParams() {
-  const headers = { "Content-Type": "application/json" };
-  if (LANGSMITH_API_KEY) {
-    headers["x-api-key"] = LANGSMITH_API_KEY;
-  }
-  return {
-    headers,
-    timeout: `${RUN_EXECUTION_TIMEOUT_SECONDS + 10}s`,
-  };
-}
-
-// Build benchmark graph options
-function buildBenchmarkGraphOptions() {
-  return {
-    graph_id: "benchmark",
-    stateful: RUN_MODE === "stateful",
-    input: {},
-    context: EFFECTIVE_CONTEXT,
-    resumable: RESUMABLE,
-  };
-}
-
-export function setup() {
-  return {};
-}
-
-export default function (data) {
-  // Print parameters (only once from VU 1)
-  if (__VU === 1 && __ITER === 0) {
-    console.log(`\n=== K6 Test Parameters ===`);
-    console.log(`BASE_URL: ${BASE_URL}`);
-    console.log(`TARGET: ${TARGET}`);
-    console.log(`BENCHMARK_TYPE: ${BENCHMARK_TYPE}`);
-    console.log(`BENCHMARK_PROFILE: ${BENCHMARK_PROFILE}`);
-    console.log(`RUN_MODE: ${RUN_MODE}`);
-    console.log(`RESUMABLE: ${RESUMABLE}`);
-    console.log(
-      `RUN_EXECUTION_TIMEOUT_SECONDS: ${RUN_EXECUTION_TIMEOUT_SECONDS}`,
-    );
-    console.log(`CONTEXT: ${JSON.stringify(EFFECTIVE_CONTEXT)}`);
-    console.log(`=========================\n`);
-  }
-
-  const startTime = Date.now();
-  const requestParams = buildRequestParams();
-  const benchmarkGraphOptions = buildBenchmarkGraphOptions();
-
-  // Get the appropriate benchmark runner
-  const runner = Benchmarks.getRunner(BENCHMARK_TYPE);
-
-  // Run the benchmark
-  const result = runner.run(BASE_URL, requestParams, benchmarkGraphOptions);
-
-  const totalDuration = (Date.now() - startTime) / 1000;
-
-  // Record execution latency (convert to ms) - do this before timeout check so we capture all durations
-  runExecutionLatency.add(totalDuration * 1000);
-
-  const timeoutSuccess = totalDuration <= RUN_EXECUTION_TIMEOUT_SECONDS;
-
-  if (!timeoutSuccess) {
-    console.log(
-      `VU ${__VU}: duration=${totalDuration.toFixed(2)}s success=false (timeout)`,
-    );
-    errorMetrics.timeout_errors.add(1);
-    fail(
-      `Benchmark execution timeout exceeded for VU ${__VU}: ${totalDuration.toFixed(2)}s > ${RUN_EXECUTION_TIMEOUT_SECONDS}s`,
-    );
-  }
-
-  // Validate the result using the runner's validation
-  const success = runner.validate(result, errorMetrics, benchmarkGraphOptions);
-
-  console.log(
-    `VU ${__VU}: duration=${totalDuration.toFixed(2)}s success=${success}`,
-  );
-
-  // Fail the VU iteration if validation failed
-  if (!success) {
-    fail(`Benchmark validation failed for VU ${__VU}`);
-  }
-}
-
-export function handleSummary(data) {
-  const latencyValues = data.metrics.run_execution_latency?.values;
-  const avgLatency = latencyValues?.avg / 1000;
-  const p95Latency = latencyValues?.["p(95)"] / 1000;
-  const p99Latency = latencyValues?.["p(99)"] / 1000;
-
-  // Collect error counts
-  const errors = {
-    server: data.metrics.server_errors?.values?.count || 0,
-    timeout: data.metrics.timeout_errors?.values?.count || 0,
-    missing_message: data.metrics.missing_message_errors?.values?.count || 0,
-    api: data.metrics.api_errors?.values?.count || 0,
-    other: data.metrics.other_errors?.values?.count || 0,
-  };
-  const totalErrors =
-    errors.server +
-    errors.timeout +
-    errors.missing_message +
-    errors.api +
-    errors.other;
-  // TODO: fix this to use actual success rate
-  const totalRequests = TARGET;
-  const successfulRequests = totalRequests - totalErrors;
-  let successRate =
-    totalRequests > 0 ? (successfulRequests / totalRequests) * 100 : 0;
-
-  return {
-    stdout: JSON.stringify({
-      target: TARGET,
-      avgExecutionLatencySeconds: avgLatency || null,
-      p95ExecutionLatencySeconds: p95Latency || null,
-      p99ExecutionLatencySeconds: p99Latency || null,
-      successRate: successRate,
-      errors: totalErrors > 0 ? errors : undefined,
-    }),
-  };
-}