Performance and Telemetry Analysis of Trae IDE: A Deep Dive into ByteDance's VSCode Fork
Executive Summary
This analysis examines concerning performance and privacy issues discovered in Trae IDE, ByteDance's fork of Visual Studio Code. Key findings include excessive resource consumption (33 processes vs 9 in VSCode), persistent telemetry transmission despite user settings, and concerning community management practices.
1. Background and Methodology
During evaluation of development environments for a personal project, I conducted a comparative analysis of three popular IDEs: Visual Studio Code, Cursor, and Trae (ByteDance's VSCode fork). This analysis revealed significant discrepancies in resource usage and network behavior that warranted deeper investigation.
Testing Environment:
OS: Microsoft Windows 11 Pro
CPU: Intel Core™ i7-14700KF
RAM: 64GB
Test Project: Identical codebase loaded in each IDE
Monitoring Tools: System Informer, Fiddler Everywhere
2. Resource Consumption Analysis
Process Count and Memory Usage
Initial testing revealed dramatic differences in resource consumption:
IDE Process Count Memory Usage Performance Impact VS Code 9 ~0.9 GB Baseline Cursor 11 ~1.9 GB 2.1x memory Trae 33 ~5.7 GB 6.3x memory
Figure 1: Trae spawns 3.7x more processes than VSCode and consumes 6.3x more memory
Community Feedback and Partial Resolution
After reporting this issue on Trae's Discord server (reference), the development team acknowledged the problem. Version 2.0.2 addressed some concerns, reducing the process count by approximately 20 processes. However, Trae still maintains significantly higher resource usage than comparable IDEs.
Post-Update Metrics (v2.0.2):
Reduced from 33 to ~13 processes
Memory usage down to ~2.5 GB
3. Network Traffic and Telemetry Investigation
Initial Discovery
Network monitoring revealed persistent outbound connections to ByteDance infrastructure:
Figure 2: Trae's network activity showing regular connections to ByteDance servers
Primary Endpoints Identified:
http://mon-va.byteoversea.com
http://maliva-mcs.byteoversea.com
https://mon-va.byteoversea.com/monitor_browser/collect/batch/?biz_id=marscode_nativeide_us
Telemetry Configuration Testing
Disabling Telemetry
I attempted to disable telemetry through the standard settings interface:
Figure 3: Telemetry disabled in user settings
Unexpected Results
Critical Finding: Disabling telemetry did not reduce network activity. Instead, it:
Maintained existing connections to mon-va.byteoversea.com and maliva-mcs.byteoversea.com
and Increased request frequency to the batch collection endpoint
Figure 4: Increase of calls
4. Data Transmission Analysis
Batch Telemetry Payload
Even with telemetry disabled, Trae continues transmitting detailed usage data:
{ "ev_type" : " batch " , "list" : [ { "ev_type" : " custom " , "payload" : { "name" : " icube_ai_ckg_request " , "type" : " event " , "metrics" : { "cost_time" : 5 }, "categories" : { "ckg_method" : " refreshToken " , "status" : " Failed " , "err_msg" : " None " , "ckg_err_code" : " SUCCEED " } }, "common" : { "bid" : " marscode_nativeide_us " , "user_id" : " redacted :) " , "device_id" : " redacted :) " , "session_id" : " redacted :) " , "env" : " prod " , "timestamp" : 1753636703370 , "sdk_version" : " 1.12.7 " , "sdk_name" : " SDK_SLARDAR_WEB " , "pid" : " AI Agent " , "view_id" : " AI Agent 1753636703370 " , "context" : { "tracing_id" : " redacted :) " , "span_id" : " 59a833c5359d4978 " , "product" : " nativeIDE " , "machine_id" : " redacted :) " , "user_id" : " redacted :) " , "user_name" : " Redacted :) " , "build_version" : " 1.0.16066 " , "app_version" : " 2.0.2 " , "build_time" : " 2025-07-21T05:08:14.915Z " , "quality" : " stable " , "eventVersion" : " 1.95 " , "arch" : " x64 " , "system" : " win32 " , "remote_arch" : " x86_64 " , "remote_system" : " windows " , "scope" : " marscode " , "biz_user_id" : " redacted :) " , "user_is_login" : " true " , "device_id" : " redacted :) " , "user_unique_id" : " redacted :) " , "organization" : " " , "vscode_version" : " 1.100.3 " , "region" : " US " , "aiRegion" : " US " , "os_name" : " windows " , "os_version" : " Microsoft Windows 11 Pro " , "cpu" : " Intel " , "is_ssh" : " false " , "app_language" : " en " , "chat_mode" : " 0 " , "identity" : " 1 " , "identity_str" : " Pro " , "pro_period" : " 0 " , "has_package" : " 0 " , "is_freshman" : " 0 " , "channel_name" : " common " , "extra" : " { \" cpu_brand \" : \" Core™ i7-14700KF \" , \" cpu_family \" : \" 6 \" , \" cpu_speed \" :3.4, \" device_manufacturer \" : \" ASUS \" , \" device_model \" : \" System Product Name \" , \" memory \" :68523634688, \" language \" : \" en-us \" } " } } } ] }
User Activity Tracking
Additional endpoint ( maliva-mcs.byteoversea.com ) receives granular user interaction data:
[ { "events" : [ { "event" : " _be_active " , "params" : " { \" start_time \" :1753636688373, \" end_time \" :1753636688373, \" url \" : \" vscode-file://vscode-app/c:/Users/segfault/AppData/Local/Programs/Trae/resources/app/out/vs/code/electron-sandbox/workbench/workbench.html \" , \" referrer \" : \"\" , \" title \" : \" read.md - tauri-app - Trae \" , \" event_index \" :1753636640233} " , "local_time_ms" : 1753636688374 , "is_bav" : 0 , "session_id" : " redacted :) " }, { "event" : " icube_time_on_ide " , "params" : " { \" entryTime \" :1753636462123, \" now \" :1753636687360, \" workspaceId \" : \" 88eda3be5f237760cfadaa7abbd89c24 \" , \" duration \" :1964, \" lastActiveTime \" :1753636685396, \" isEditorFocused \" :false, \" activeEditorTypeId \" : \" workbench.editors.files.fileEditorInput \" , \" isMouseOrKeyboardActive \" :true, \" isAiActive \" :false, \" activeEditorProviderId \" : \"\" , \" activeEditorLanguage \" : \" markdown \" , \" activeEditorFileExt \" : \" .md \" , \" isFocus \" : \" not_focus \" , \" isVisible \" : \" hidden \" , \" event_index \" :1753636640232} " , "local_time_ms" : 1753636688372 , "is_bav" : 0 , "session_id" : " redacted :) " } ], "user" : { "user_unique_id" : " redacted :) " , "user_id" : " redacted :) " , "user_is_login" : true , "device_id" : " redacted :) " }, "header" : { "app_id" : 677332 , "app_version" : " 2.0.2 " , "os_name" : " windows " , "os_version" : " Microsoft Windows 11 Pro " , "device_model" : " System Product Name " , "language" : " en-us " , "region" : " US " , "app_language" : " en " , "platform" : " electron " , "sdk_version" : " 5.1.25 " , "sdk_lib" : " js " , "timezone" : 2 , "tz_offset" : -7200 , "resolution" : " 2560x1440 " , "browser" : " Chrome " , "browser_version" : " 132.0.6834.210 " , "referrer" : " " , "referrer_host" : " " , "width" : 2560 , "height" : 1440 , "screen_width" : 2560 , "screen_height" : 1440 , "tracer_data" : " { \" $utm_from_url \" :1} " , "custom": "{\"icube_uid\":\"7472574750745953285\",\"biz_user_id\":\"7472574750745953285\",\"is_special_uuid\":false,\"machine_id\":\"9531bb512f2c1aff9c2ed17dd9e342026f0e1e78326b35ee01e03f6439b869ca\",\"arch\":\"x64\",\"system\":\"win32\",\"scope\":\"marscode\",\"organization\":\"\",\"build_version\":\"1.0.16066\",\"vscode_version\":\"1.100.3\",\"tenant\":\"marscode\",\"aiRegion\":\"US\",\"quality\":\"stable\",\"build_time\":\"2025-07-21T05:08:14.915Z\",\"icube_main_uid\":\"f8207401-9c52-4583-82d3-e1e44e2da0f0\",\"window_id\":2,\"workspace_id\":\"88eda3be5f237760cfadaa7abbd89c24\",\"os_release\":\"10.0.26100\",\"os_build\":\"26100\",\"device_manufacturer\":\"ASUS\",\"cpu\":\"Intel\",\"cpu_brand\":\"Core™ i7-14700KF\",\"cpu_vendor\":\"GenuineIntel\",\"cpu_family\":\"6\",\"cpu_model\":\"183\",\"cpu_stepping\":\"1\",\"cpu_speed\":3.4,\"memory\":68523634688,\"is_ssh\":false,\"chat_mode\":0,\"identity\":\"1\",\"identity_str\":\"Pro\",\"pro_period\":\"0\",\"has_package\":\"0\",\"is_freshman\":\"0\",\"channel_name\":\"common\"}" }, "local_time" : 1753636689 , "verbose" : 1 } ]
Data Collection Scope
The telemetry system captures:
System Information : Hardware specs, OS details, architecture
: Hardware specs, OS details, architecture Usage Patterns : Active time, session duration, feature usage
: Active time, session duration, feature usage Performance Metrics : Response times, resource consumption
: Response times, resource consumption Unique Identifiers : Machine ID, user ID, device fingerprints
: Machine ID, user ID, device fingerprints Workspace Details: Project information, file paths (obfuscated)
👉 Watch the Network Calls on Streamable
5. Community Management Concerns
Automated Censorship
When I attempted to discuss these findings on Trae's Discord server, i got spanked with gag-hammer https://discord.com/channels/1320998163615846420/1335032920850825391/1398374824987852891
The moderation blacklist was added after the discussion taken place, the mute was manual at first.
Keyword Filtering: The word "track" was added to an automated blacklist Automatic Punishment: Mentioning tracking issues triggered an instant 7-day mute Suppression of Technical Discussion: Legitimate security concerns were treated as disruptive behavior
6. Privacy and Security Implications
Data Sovereignty Concerns
Persistent Collection : Telemetry continues despite user preferences
: Telemetry continues despite user preferences Granular Tracking : Detailed system and usage information transmitted
: Detailed system and usage information transmitted Foreign Data Processing : Information routed to ByteDance (Chinese company) infrastructure
: Information routed to ByteDance (Chinese company) infrastructure Unique Identification: Multiple persistent identifiers enable long-term tracking
Trust and Transparency Issues
Misleading Settings : Telemetry toggle appears non-functional
: Telemetry toggle appears non-functional Undocumented Behavior : No clear disclosure of data collection practices
: No clear disclosure of data collection practices Community Suppression: Technical criticism met with censorship rather than engagement
Key Takeaways:
Resource usage is 6x higher than VSCode baseline
Telemetry settings appear to be cosmetic rather than functional
Community feedback mechanisms are compromised by censorship
Data collection practices lack transparency and user control
This analysis was conducted in July 2025 using Trae IDE version PRE-2.0.2 and 2.0.2. Network traffic was captured using standard monitoring tools, and all findings are reproducible. Community members are encouraged to conduct their own testing and share results through appropriate channels.
While core of this research was written by hand, an LLM has rewritten it to fix broken english, grammar and enhance the verbal aspect of research.
Discord: cryptux x (twitter): https://x.com/CookingCodes