Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add kevinpbuckley/VibeUE --skill profilinggit clone --depth 1 https://github.com/kevinpbuckley/VibeUEWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/kevinpbuckley/vibeue/profiling)<a href="https://agentmods.dev/skills/kevinpbuckley/vibeue/profiling"><img src="https://agentmods.dev/badge/skills/kevinpbuckley/vibeue/profiling/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/kevinpbuckley/vibeue/profiling"><img src="https://agentmods.dev/badge/skills/kevinpbuckley/vibeue/profiling.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00057 | $0.04519 |
| Opus 5.5 | $0.00023 | $0.01808 |
| Sonnet 5.5 | $0.00011 | $0.00904 |
| Haiku 4.5 | $0.00006 | $0.00452 |
Grade A, and why
profiling scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 366 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Profiling Skill
Unreal 5.8's native toolsets have no performance or tracing tools, so this is VibeUE's net-new
capability: unreal.PerformanceService answers "are we CPU- or GPU-bound?" and drives Unreal
Insights captures from Python with no C++ required. Run everything here through execute_python_code.
The whole API is small — read the live signatures once with
discover_python_class(class_name="unreal.PerformanceService"):
| Method | Purpose |
|---|---|
frame_timing(target_fps=60.0) |
Game/Render/GPU/RHI thread ms + a CPU-vs-GPU bound verdict + a hint, plus a per-thread budget breakdown gated against target_fps (budget.meets_target PASS/FAIL). Run FIRST. |
force_hitch(thread="game", ms=250.0, frames=1) |
Validation helper — deliberately stall a KNOWN thread ("game"/"render"/"both"/"gpu") to confirm the verdict fires. CPU paths are self-measuring: validate from the SAME response (observed_peak_*_ms, verdict_matched_expect) — do NOT follow up with frame_timing. Only the async gpu path is read back on a following frame. Clamped (≤5000 ms, ~10 s total) for safety. |
report(title=..., source="both", file="") |
Write a self-contained, shareable HTML report (live verdict + budget + analyse stats + a data-driven "fix in this order" list) to Saved/VibeUE/Performance/report_<timestamp>.html. |
start_pie() / stop_pie() |
In-process PIE so frame_timing/force_hitch read a live game world (PIE starts on the next tick — read on a following frame, pie_running flips true). Quick checks only; prefer start_standalone when the numbers must be trusted. |
start_trace(name="mcp_capture", channels="") |
Start an Insights trace to file (default channel set if channels empty). |
stop_trace() |
Stop the active trace; returns the trace file path + size. |
get_trace_status() |
Whether a trace is active and which channels are enabled. |
bookmark(name) |
Drop a point-in-time bookmark in the active trace. |
region_start(name) / region_end(name) |
Begin / end a named region span in the active trace. |
analyse(source="both", file="") |
Read back trace and/or log → frame stats, worst frames, hitches, notable log lines. Combined analysis reports complete, partial, or failed; partial is never top-level success. |
start_standalone(name="standalone_capture", channels="") |
Request a separate process with a unique direct-to-file trace. The first response is start_pending; poll status for verified capture. |
stop_standalone() |
Gracefully stop the exact tracked process and verify the exact trace finalized; forced/partial outcomes are explicit failures. |
get_standalone_status() |
Session ID, PID, map, exact trace/log destinations, process state, and capture/finalization verification. |
All methods return a JSON string. For a representative reading, profile under PIE or a standalone
session (start_standalone), not the bare editor viewport.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago Changed · +19 lines fab1f25744db
- 1mo ago First seen · 347 lines · 57 tokens per session scan A 4cab72ae93c1
profiling is a skill published in the GitHub repository kevinpbuckley/VibeUE (787 stars, last pushed yesterday), licensed MIT. It adds 57 tokens to every session and 4,519 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
cli-anything-nsight-graphics
Windows-first CLI harness for Nsight Graphics capture, GPU Trace summary, and ngfx-replay analysis.
cuda-kernel-optimizer
Use when optimizing or diagnosing GPU workload performance with CUDA, CUTLASS, Triton, PyTorch, vLLM or TensorRT-LLM; interpreting NCU, Nsys or PyTorch Profiler evidence; or preparing the workload and correctness checks for such optimization. General coding, skill maintenance and conceptual GPU questions do not…
qa-test
Run comprehensive QA tests for SpikeTrap. Covers all API methods, collect mode lifecycle, filter accuracy, domain reload survival, edge cases, and UI verification. Use when testing after code changes or before release.
profiling
A set of Linux instructions for finding which parts of a running program or system use CPU time, memory, or other resources.
spike-trap
Run SpikeTrap CPU profiling workflow. Use to: (1) Profile the game and capture spike frames, (2) Analyze existing cached profiler data for bottlenecks, (3) Save matched frames to .data files. Requires Unity Profiler window open with SpikeTrap module selected.
performance-optimizer
A performance improvement assistant that analyses applications, identifies bottlenecks, and suggests ways to make them faster.