agentleFS
hass-energyhass-energy/haeo/.agents/skills/performance/SKILL.md

performance

Skill77 starsChanged 20 days agoMITView on GitHub

Measuring and improving HAEO's speed — optimization cycle time, input loading on the event loop, LP rebuild and solve time, output sensor building, pytest-benchmark model benchmarks, and profiling with cProfile. Use when asked to make something faster, when a change could affect solve or load time, when Home Assistant reports a slow or blocking update, or before claiming a performance improvement.

86 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use app-brief, design-from-app and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • StijnusSkill

    app-brief

    One-page brief (docs/BRIEF.md) for a new macOS app, written before any code. Use when the user has an app idea, wants to replace an app they pay for, or the research and clone skills are done.

    Also in 3 projects that use this

  • StijnusSkill

    design-from-app

    Rebuild the look of an app the user likes as a native SwiftUI design kit, by measuring its screenshots (sizes, spacing, type, colors, icons), building its parts, and comparing pixel by pixel until they match. Use when the user names an app whose look they want, shares screenshots of one, or says "make it look like X".

    Also in 3 projects that use this

  • FlorianBruniauxSkill

    tdd-rust

    Also in 3 projects that use this

  • RodrigoTomeESSkill

    web-quality-audit

    Comprehensive web quality audit covering performance, accessibility, SEO, and best practices. Use when asked to "audit my site", "review web quality", "run lighthouse audit", "check page quality", or "optimize my website".

    Also in 9 projects that use this

  • launchapp-devSkill

    layout-grid

    Define responsive layout grid systems with columns, gutters, margins, and breakpoint behavior.

    Also in 3 projects that use this

  • MohamedAbdallah-14Skill

    app-icon

    Generate an app icon from a single 1024² master and fan out to iOS (AppIcon.appiconset with squircle-ready 1024 opaque), Android (adaptive foreground + background + Android 13 monochrome), PWA (192/512 + 512 maskable with 80% safe zone), and visionOS (3-layer parallax).

    Also in 3 projects that use this

  • +1 more from Stijnus

The file

We split this skill into the rules it gives an agent

3 rules. 0 are common, 2 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 58 edited versions we read keep these.

  • 2

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Model rebuild and solve: run in the executor. The network persists between cycles: each horizon roll changes the period widths, which …

    line 14
  2. Distinctive:

    Output extraction and sensor building: run on the event loop after every solve.

    line 16
  3. Very specific:

    Run uv run pytest tests/scenarios/test_benchmark.py -m benchmark --benchmark-enable and compare medians against the base branch.

    line 32

“Common” is measured: kept by at least 6 in 10 of the 58 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Performance work

Stages of an optimization cycle

Measure each stage before choosing what to change; which one dominates depends on the configuration and the path being exercised.

  • Input loading runs on the Home Assistant event loop whenever a source entity changes. Forecast integrations write many entities in one pass, so per-change work multiplies by the number of sources in a field.
  • Model rebuild and solve run in the executor. The network persists between cycles: each horizon roll changes the period widths, which rebuilds most constraints, and HiGHS then warm starts.
  • Output extraction and sensor building run on the event loop after every solve.
  • Cold start (network build plus the first solve) happens only at setup or reload.

Rules

  • Replace per-period, per-point, and per-row Python loops with numpy operations on whole arrays. In the model layer this means LinearArray; the model skill covers how.
  • Work done for every element, output, or sensor must not repeat work that is the same for all of them, such as converting horizon timestamps.
  • Coalesce bursts of state changes into one reload, but load immediately when an optimization could otherwise read stale inputs.
  • A performance change must not change results. Dump every model output before and after across cold solves and horizon rolls and compare them. Costs and primal outputs must match; shadow prices may differ only where the LP is degenerate, and the cause must be explained.

Measuring

  • Model benchmarks are disabled by default. Run uv run pytest tests/scenarios/test_benchmark.py -m benchmark --benchmark-enable and compare medians against the base branch.
  • A benchmark round must change something real. Re-solving unchanged inputs measures a warm no-op, which is why test_time_shift alternates between two horizons and updates the periods.
  • The helpers in tests/scenarios/test_benchmark.py build a network from a scenario without Home Assistant, which suits ad hoc timing and cProfile scripts.
  • Split solve time from Python time by timing the solver run() calls inside Network.optimize() (the Highs subclass in core/model/solver.py); the remainder is model rebuild.

Profiling pitfalls

  • Scenario tests run under freezegun, which also freezes time.perf_counter. Time with freezegun.api.real_perf_counter, and expect datetime-heavy code to look several times slower than it is in production.
  • On Python 3.12 and later, cProfile records every thread, so executor work appears in an event loop profile. Its timings include waiting for the GIL.
  • Profile the path a user actually hits, such as a burst of source entity changes or a horizon roll, not only a forced re-run with unchanged inputs.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.