· 1 observation
Technical updates · 1
Package or technical metadata updated
View technical comparison →TimmyTook v0.3.2
Analyze Tarkov performance
Developer declarations; not independently tested. Listing last checked Nov 4, 2026. Package observed Nov 3, 2026.
Not specified in the available declarations: limitations, pricing & account requirements.
Analyze this Tarkov Performance Toolkit report.
From the marketplace listing
Analyze Escape from Tarkov settings and user-provided performance data with four read-only workflows for configuration review, FPS and frametime interpretation, repeatable benchmark comparison, and controlled tuning. Intended for players diagnosing frame-rate, stutter, and stability problems. ChatGPT cannot access the user's PC; users explicitly paste or attach sanitized Toolkit output or compatible benchmark data. The plugin does not edit game files, automate gameplay, read process memory, or upload benchmark data.
Language: English · Automatically detected from descriptions.
· 1 observation
Package or technical metadata updated
View technical comparison →Find competitors to follow as you develop your plugin.
AI-native, governance-first SQL Server and PostgreSQL orchestration plugin.
Shared terms: performance, analyze, controlled, repeatable
Ask ChatGPT about the events your Checkin account can access—without exporting reports. Find events, attendees, contacts, and orders; mon...
Shared terms: performance, analyze, review, read
Hologrow makes your connected commerce, advertising, analytics, and ERP data available for analysis. Explore performance, compare channel...
Shared terms: analyze, analysis, performance, setting
Review connected Google Ads and Meta Ads campaigns. Find workspaces and ad accounts, inspect campaign settings, and compare spend, clicks...
Shared terms: performance, cannot, setting, review
Connect Google Search Console to analyze finalized search performance, uncover keyword and content opportunities, compare ranking changes...
Shared terms: analyze, cannot, performance, edit
--- name: tarkov-config description: Analyze Escape from Tarkov graphics configuration and Windows performance context, or guide optional first-time Game-tab setup, without changing game files. Use for FPS expectations, stutter risks, visibility tradeoffs, or a saved performance goal; prefer the signed Toolkit for collection and support transparent manual review when it is unavailable. --- # Tarkov Config Analyze settings and system readiness in read-only mode. Never edit EFT files or promise a specific FPS. ## Optional First-Time Setup For first-time setup or a request for beginner Game settings, use `references/game-starting-profile.md`. Offer its manual Game-tab profile separately from FPS optimization; preserve the user's preferences and skip settings already matching. Do not proactively repeat it during ordinary FPS troubleshooting or after the user accepted/declined it in the conversation; revisit it only if requested. A setup-only request needs no Toolkit report or capture. ## Choose The Input Path 1. For a local agent, run `tarkov-skills.exe status`. If available, run `tarkov-skills.exe inspect` and consume its JSON. 2. In a web client, ask the user to open **Tarkov Performance Toolkit**, press **Collect report**, then **Copy JSON** and paste or attach the result. 3. Without Toolkit, explain each source before reading it. A local agent may directly read `Graphics.ini`, `PostFx.ini`, and `Game.ini` and use standard read-only Windows system queries. A web user may attach those files and a Windows System Information report. Do not download, generate, or execute script files. The Toolkit is the only automated dependency. If it is missing, direct the user to `https://apps.microsoft.com/detail/9N3L7DZH0K64`; do not offer unsigned PowerShell as a fallback. ## Goal Read the active goal from `inspect` or `tarkov-skills.exe goal get`. When the user changes the target or quality tradeoff, confirm the values and save them with: ```text tarkov-skills.exe goal set --goal <name> --target-fps <20-360> --quality <text> [--notes <text>] ``` In web/manual mode, keep the goal in the conversation when the local command cannot be run. ## Analysis - For performance analysis, use only `Graphics.ini`, `PostFx.ini`, and graphically relevant `Game.ini` data. The optional Game-tab profile is separate manual guidance; ignore Controls/Sound settings in both paths. - Consider CPU, GPU/VRAM, RAM, game-drive media, pagefile size/media, resolution, and driver version. - Suggest only changes that are not already applied. - If measured performance is at least 15% below a relevant expectation, switch from ordinary graphics tuning to diagnostics such as throttling, power, RAM configuration, storage, drivers, overlays, and local PvE load. - Treat driver currency as a manual check against the GPU vendor's official page. Use the readiness display and response conventions in `references/configuration-rules.md`. Apply thresholds from `references/measurement-rules.md`.
Referenced files: 4 — file references are metadata, not package contents.
--- name: tarkov-frametime description: Collect or interpret read-only Escape from Tarkov FPS and frametime metrics. Use for average FPS, 1% low, 0.1% low, p95/p99 frametime, duration, and sample count; prefer the signed Tarkov Performance Toolkit and leave map/settings interpretation to benchmark or tuning skills. --- # Tarkov Frametime Collect performance metrics without changing game files, reading game memory, injecting code, or automating input. ## Automated Capture For a local agent: 1. Run `tarkov-skills.exe status`. 2. Confirm that Tarkov is running and `raid_active` is true. 3. Tell the user that capture uses bundled PresentMon and ask before starting the timed measurement. 4. Run `tarkov-skills.exe capture --duration 120`, or `240` when requested. 5. Consume the returned JSON. A nonzero exit and JSON status such as `permission_required`, `capture_conflict`, or `failed` is a failed capture; do not keep partial results. For a web client, ask the user to open **Tarkov Performance Toolkit**, select **Benchmark**, enter a raid, press **Start collection**, save the completed run, then press **Copy results** and paste the JSON into the conversation. If Toolkit is unavailable, direct the user to `https://apps.microsoft.com/detail/9N3L7DZH0K64`. There is no unsigned-script capture fallback. You may interpret an existing PresentMon, CapFrameX, or FrameView CSV the user already has, but do not claim an automated capture occurred. Return average FPS, 1% low, 0.1% low, p95/p99 frametime, duration, sample count, and method. Do not collect map, weather, server execution, or settings conclusions in this skill.
Referenced files: 2 — file references are metadata, not package contents.
--- name: tarkov-performance-benchmark description: Guide a repeatable read-only Escape from Tarkov performance benchmark with current settings, system context, raid logs, and FPS/frametime metrics. Use the signed Store applications for automated collection and produce a normalized run without modifying game files. --- # Tarkov Performance Benchmark Record a benchmark; do not optimize settings in this skill. ## Preferred Store Flow For a local agent, use the signed Toolkit without opening its GUI: 1. Run `tarkov-skills.exe status` and require an active raid. 2. Ask before starting capture, then run `tarkov-skills.exe capture --duration 120` or `240` when requested. 3. Use the report's settings, system, map/log context, and performance metrics. 4. Ask only for missing `BSG servers` versus `Local`, weather, and time of day. Server execution is required; weather/time may be `unknown`. Do not open the GUI when the local alias is callable. For a web client, ask the user to open **Tarkov Performance Toolkit**, select **Benchmark**, complete and save the run, press **Copy results**, and paste the JSON into the conversation. `Copy results` copies only the latest complete run and uploads nothing. The Toolkit Store page is `https://apps.microsoft.com/detail/9N3L7DZH0K64`. Without Toolkit, accept manually attached settings plus an existing capture export, but do not provide or execute unsigned scripts. ## Rules - Start only when Tarkov is running and logs show an active raid. - Default to 120 seconds; allow 240 seconds on request. - Prefer map and game version from logs. Ask for the map only if log recovery fails. - Do not ask for route, activity, PvP/PvE, or a separate server-model field. - Discard interrupted, cancelled, game-crash, or raid-exit captures. - Keep settings and system context with each run used for comparison. - Do not include usernames, hostnames, local paths, IPs, serial numbers, or machine identifiers. - Do not recommend graphics changes here. Use `tarkov-tuning` for that. Apply confidence and repeatability rules from `references/measurement-rules.md`.
Referenced files: 4 — file references are metadata, not package contents.
--- name: tarkov-tuning description: Orchestrate iterative Escape from Tarkov performance tuning from read-only configuration and measured frametime data. Use the signed Tarkov Performance Toolkit, preserve the user's FPS/quality goal, recommend manual changes, and compare repeatable before/after captures. --- # Tarkov Tuning Coordinate `tarkov-config` and `tarkov-frametime`; use `tarkov-performance-benchmark` when full run context matters. Never edit EFT settings automatically. ## Optional Setup Before The Baseline When the user is setting up Tarkov for the first time or requests beginner Game settings, offer `references/game-starting-profile.md` before the first baseline. Reuse any choice already handled by `tarkov-config` in the conversation. It is manual usability guidance, not an FPS change batch; declining it does not block tuning. Skip it for an established setup and do not repeat it during the loop unless the user asks to revisit it. ## Loop 1. Run `tarkov-skills.exe inspect` and read the saved goal. In a web client, ask the user to use **Overview → Collect report → Copy JSON** and paste it into the conversation. 2. If the user changes the target or quality tradeoff, save it through `tarkov-skills.exe goal set ...`; in web/manual mode, retain it in the conversation. 3. Measure a baseline with Toolkit after explicit confirmation. In a web client, guide the user through **Benchmark → Start collection → Save benchmark → Copy results** and read the pasted latest-run JSON. 4. Recommend one small manual setting group that is not already applied. 5. Repeat the same duration and similar scenario. 6. Keep, revert, repeat, or switch to diagnostics using `references/measurement-rules.md`. If Toolkit is missing, direct the user to `https://apps.microsoft.com/detail/9N3L7DZH0K64`. Do not download, create, or execute unsigned scripts. Transparent manual config review is allowed, and an existing CSV may be interpreted, but automated frametime capture requires Toolkit. Prioritize 1% low for stability goals and average FPS for maximum-FPS goals. For quality goals, accept lower FPS only while the user's target remains satisfied. Results inside the noise threshold require another run; a miss of at least 15% against a relevant expectation triggers diagnostics. Use the output format in `references/tuning-loop.md`.
Referenced files: 4 — file references are metadata, not package contents.
Publisher declarations from the archived package. These are separate from our research and the live service's terms.
Package observed Nov 3, 2026.
plugins_6ac208ab1c0c8191beae4140c8d7ceb7
Download plugin data (JSON)