desktop-app-audit

星标4

分支0

更新时间2026年7月18日 15:25

Audit desktop applications — especially Electron and other web-runtime apps — for idle resource waste: background CPU when unfocused, memory footprint, startup daemons, per-app runtime duplication, animation and polling loops that never sleep. Use this skill whenever the user builds or maintains a desktop app, mentions Electron/Tauri/WebView apps, shares process or profiler data, or users complain the app drains laptops. Part of Lean Agentic AI Skills; emits lean-findings.json.

安装

用 Codex 或 Claude 帮你安装复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。

来源

navveenb

navveenb/lean-agentic-ai-skills

打开 GitHub 仓库查看创作者相关仓库

下载

Desktop App Audit

Subject type: emit subject.type: "desktop-app" in findings.

Producer skill. Input: app architecture description, profiler/process data (CPU when idle/unfocused, RSS), update and background-service configs, renderer counts. Output: lean-findings.json.

Desktop apps run for hours on hardware the user owns — idle CPU is battery drained and fans spun across every installed seat, and memory pressure shortens the useful life of machines (the M-lever lean-mobile-audit's radio logic doesn't cover). Web-runtime apps are the biggest offenders by construction: each ships its own browser.

Signatures

Busy while idle — measurable CPU when the window is unfocused/minimized: animation loops not paused, timers firing, telemetry heartbeats. High (hours × seats). Evidence: profiler CPU% in idle state. Fix: pause rendering on blur/occlusion (Electron: backgroundThrottling, visibility handlers), coalesce timers.
Polling where OS events serve — file watchers implemented as poll loops, network polling for state push could deliver. High. Fix: native FS events, push channels, OS-scheduled background tasks.
Startup daemons and helpers — auto-start services running always for features used rarely (updaters resident 24/7 to check once a day). Medium-high. Fix: on-demand launch, OS-scheduled checks.
Renderer sprawl — one process/webview per panel; memory multiplying with windows. Medium-high. Evidence: process count × RSS. Fix: consolidate renderers, unload hidden views.
Web runtime where native serves — the honest architecture finding: a status-bar utility shipping a full browser runtime. Medium; effort high; alternatives (Tauri/native webview, native UI) named with their trade-offs (team skills, cross-platform cost) — this is an option, never a mandate.
Update churn — full-app downloads for small updates; no delta updates. Medium: bytes × seats × frequency. Fix: differential updates.
Unbounded caches/logs on user disks — app data growing forever on machines you don't pay for. Low-medium. Fix: cache ceilings, log rotation.
Ignoring OS power states — same behavior on battery vs plugged, no response to low-power mode. Medium. Fix: honor power APIs; heavy work deferred to plugged+idle (the desktop cousin of carbon-aware scheduling).

Cost signal (countable)

Provider-side cost is mostly update bandwidth (GB × seats × releases) and any telemetry ingestion — countable from release sizes and analytics volume. The dominant costs are user-side (battery, machine longevity): name them in impact text, not cost_signal, per the module's rule against charging users' costs to fabricated figures.

Honesty rules

Idle-CPU claims come from a profiler run, not intuition — name the tool (OS activity monitors, Electron's contentTracing, platform profilers). Runtime-replacement recommendations always carry the engineering-cost trade-off at effort: high. Seat counts are the user's data.

Not this skill's job

Web pages in browsers (web-carbon-audit), the app's bundle build (js-bundle-audit — Electron apps should run both), mobile (lean-mobile-audit), backend APIs it calls (api-efficiency-audit).

同仓库更多 Skills

同仓库

agentic-loop-auditor

navveenb/lean-agentic-ai-skills

Audit agent frameworks and agentic workflows for token-burning loop pathologies — unbounded retries, excessive planning depth, tool-call thrashing, context snowballing, missing loop budgets and kill switches. Use this skill whenever the user shares agent configs or traces (LangChain/LangGraph, CrewAI, AutoGen, custom loops), complains an agent is slow or expensive, or is designing any autonomous/multi-step agent. Part of Lean Agentic AI Skills; emits lean-findings.json.

2026-07-184

ai-eval-efficiency-audit

navveenb/lean-agentic-ai-skills

Audit AI evaluation infrastructure for wasted compute — full benchmark suites re-run on unchanged cases, LLM-as-judge grading without caching, oversized judge models, redundant eval passes per commit, and missing result reuse. Use this skill whenever the user shares eval harness configs or CI eval steps, complains that evals are slow or expensive, mentions LLM-as-judge costs, or runs benchmark suites on every change. Part of Lean Agentic AI Skills; emits lean-findings.json.

2026-07-184

api-efficiency-audit

navveenb/lean-agentic-ai-skills

Audit API design for chattiness and over-fetching — REST endpoints requiring N calls per screen, GraphQL resolvers with N+1 fan-out, default-everything payloads, missing pagination and field selection, provider-side polling where webhooks serve, and uncompressed responses. Use this skill whenever the user shares API schemas (OpenAPI, GraphQL SDL), resolver code, endpoint designs, or asks why clients make too many calls or payloads are heavy. Part of Lean Agentic AI Skills; emits lean-findings.json.

2026-07-184

caching-strategy-designer

navveenb/lean-agentic-ai-skills

Design caching layers so work is done once, not per request — HTTP/CDN caching, application-level caches (Redis/Memcached), computed-result caching, and cache-invalidation strategy. Use this skill whenever the user wants to add caching, asks why the same work is recomputed, designs a read-heavy system, or an audit skill (db-efficiency-audit, web-carbon-audit) flags uncached repeated work. Part of Lean Agentic AI Skills; emits a cache design plus lean-findings.json.

2026-07-184

carbon-aware-scheduler

navveenb/lean-agentic-ai-skills

Design time-shifting of flexible workloads (batch jobs, CI, training, backups, crons) to hours and regions when grid electricity is cleanest, using carbon-intensity signals like Electricity Maps, WattTime, or the Carbon Aware SDK. Use this skill whenever the user mentions carbon-aware computing, scheduling flexible/deferrable jobs, demand shifting, running workloads when energy is clean, or asks how to reduce the carbon intensity (not just amount) of their compute. Part of Lean Agentic AI Skills; emits schedule designs plus lean-findings.json.

2026-07-184

carbon-budget-designer

navveenb/lean-agentic-ai-skills

Design per-feature and per-service efficiency budgets — page-weight budgets, token budgets per AI feature, query budgets per endpoint, GPU-hour budgets per training project — with owners, review checkpoints, and enforcement hooks. Use this skill whenever the user wants sustainability targets for teams, asks how to keep efficiency from regressing organizationally, mentions performance-budget-style governance for carbon/energy, or is setting engineering standards. Part of Lean Agentic AI Skills; an advisor emitting budget definitions plus lean-findings.json.

2026-07-184

name

desktop-app-audit

description

Desktop App Audit

Subject type: emit subject.type: "desktop-app" in findings.

Producer skill. Input: app architecture description, profiler/process data (CPU when idle/unfocused, RSS), update and background-service configs, renderer counts. Output: lean-findings.json.

Signatures

Busy while idle — measurable CPU when the window is unfocused/minimized: animation loops not paused, timers firing, telemetry heartbeats. High (hours × seats). Evidence: profiler CPU% in idle state. Fix: pause rendering on blur/occlusion (Electron: backgroundThrottling, visibility handlers), coalesce timers.
Polling where OS events serve — file watchers implemented as poll loops, network polling for state push could deliver. High. Fix: native FS events, push channels, OS-scheduled background tasks.
Startup daemons and helpers — auto-start services running always for features used rarely (updaters resident 24/7 to check once a day). Medium-high. Fix: on-demand launch, OS-scheduled checks.
Renderer sprawl — one process/webview per panel; memory multiplying with windows. Medium-high. Evidence: process count × RSS. Fix: consolidate renderers, unload hidden views.
Web runtime where native serves — the honest architecture finding: a status-bar utility shipping a full browser runtime. Medium; effort high; alternatives (Tauri/native webview, native UI) named with their trade-offs (team skills, cross-platform cost) — this is an option, never a mandate.
Update churn — full-app downloads for small updates; no delta updates. Medium: bytes × seats × frequency. Fix: differential updates.
Unbounded caches/logs on user disks — app data growing forever on machines you don't pay for. Low-medium. Fix: cache ceilings, log rotation.
Ignoring OS power states — same behavior on battery vs plugged, no response to low-power mode. Medium. Fix: honor power APIs; heavy work deferred to plugged+idle (the desktop cousin of carbon-aware scheduling).

Cost signal (countable)

Honesty rules

Not this skill's job

Web pages in browsers (web-carbon-audit), the app's bundle build (js-bundle-audit — Electron apps should run both), mobile (lean-mobile-audit), backend APIs it calls (api-efficiency-audit).