SECURITY REVIEW
Not yet assessed
Review the original instructions and requested permissions before installing.
No security review is available for this catalog entry yet.
Community / AGENT SKILL
performance-analyzer
0
Read-only performance analysis for Blunder Engine hot paths. Use with Task tool for parallel review of viewport, render, or UI bottlenecks. Reports bottlenecks with severity and suggested fix order.
Understand the trade-offs.
SKILL QUALITY
Not yet assessed
How clearly the skill guides your agent, how complete its workflow is, and how you can check the outcome.
No quality assessment is available for this catalog entry yet.
The full skill.
Original instructions from the publisher’s SKILL.md
# Performance analyzer (read-only) Run as a **parallel subagent** (Cursor Task tool, `subagent_type: explore`, `readonly: true`). ## Scope - Viewport / render pipeline (`RenderSystem`, readback, Slint present) - Per-frame allocations and redundant work - GPU/CPU sync stalls (fences, readback, full-window composite) Do **not** edit files. Read and report only. ## Process 1. Read `docs/agents/render-pipeline.md` and trace the reported hot path. 2. Grep/read key files under `engine/src/runtime/function/render/` and `.../ui/`. 3. Identify: - Work done every frame vs only on dirty/resize - Full-screen vs partial updates - CPU-GPU stalls and copies - Redundant Slint or Vulkan passes 4. Rank findings by **impact** (high / medium / low) and **effort** (small / medium / large). ## Output format ```markdown ## Performance findings ### High impact - [finding] — file:line — suggested fix ### Medium impact - ... ### Low impact / nice-to-have - ... ## Recommended order 1. ... 2. ... ## Risks - Invariants that must not break (Z-up, offscreen path, Slint fork) ``` ## Pair with - `viewport-perf` skill for viewport-specific context - `blunder-engine` skill for project invariants