What is speedup_native?

A workflow for finding and fixing performance problems in Kotlin/Native, Kotlin's compiler and runtime for native platforms. It compares the same test on JVM and Native targets, measures each phase, and applies small platform-specific changes.

Use it when Kotlin/Native code is slower than its JVM version, especially in Kotlin Multiplatform projects, to reproduce the issue, profile it, and make minimal expect/actual optimizations.

It replaces guesswork with repeatable measurements and helps isolate the slowest step before code is changed.

How nehemiaharchives/lucene-kmp configures Claude Code and Codex

Reuse

Borrowing it

Nothing to install: this file belongs to nehemiaharchives/lucene-kmp. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.

Copy the file

curl -O https://raw.githubusercontent.com/nehemiaharchives/lucene-kmp/master/.agents/skills/speedup_native/SKILL.md

Clone the repo

git clone --depth 1 https://github.com/nehemiaharchives/lucene-kmp

Made for: Claude Code, Codex.

Wrote 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.

README.md

[![agentmods](https://agentmods.dev/badge/skills/nehemiaharchives/lucene-kmp/speedup_native/github.svg)](https://agentmods.dev/skills/nehemiaharchives/lucene-kmp/speedup_native)

Your own site

<a href="https://agentmods.dev/skills/nehemiaharchives/lucene-kmp/speedup_native"><img src="https://agentmods.dev/badge/skills/nehemiaharchives/lucene-kmp/speedup_native/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.

Your own site · 80×15

<a href="https://agentmods.dev/skills/nehemiaharchives/lucene-kmp/speedup_native"><img src="https://agentmods.dev/badge/skills/nehemiaharchives/lucene-kmp/speedup_native.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 32 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.

When invoked 1,002 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

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	$0.00032	$0.01002
Opus 5.5	$0.00013	$0.00401
Sonnet 5.5	$0.00006	$0.00200
Haiku 4.5	$0.00003	$0.00100

Measured 16d ago against content hash 8863a600d5dc, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

speedup_native 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 16d 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.

.agents/skills/speedup_native/SKILL.md · 111 lines

How it starts

The opening of the file, as written. The whole thing — 111 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Purpose

Use this skill when a test or code path is significantly slower on Kotlin/Native than JVM and you need a reliable, repeatable way to find and fix the lowest-level bottleneck.

Scope

Kotlin Multiplatform projects with JVM + Native targets.
Works best when one target is clearly slower (for example, linuxX64Test vs jvmTest).
Focuses on minimal-risk optimizations that preserve behavior.

Core Principles

Reproduce first, optimize second.
Measure with monotonic clocks, not assumptions.
Always isolate the lowest-level slow step before patching.
Keep patches minimal and reversible.
Prefer compile/test truth over IDE false negatives for expect/actual wiring.

End-to-End Workflow

1. Reproduce with smallest failing/slow test

Run exactly one test method that exhibits slowness.
Keep both target runs comparable (same method, same data).

2. Add phase-level timing in test (coarse)

Use TimeSource.Monotonic.markNow().
Add logs for major phases only first.
Include total elapsed and per-phase elapsed.

3. Compare JVM vs Native and pick worst phase

Identify the phase with the largest native/JVM ratio.
Ignore small differences; chase the dominant hotspot.

4. Add substep timing inside worst phase (fine)

Split compound calls into substeps (a().b() -> time a and b separately).
Iterate until one substep clearly dominates native time.

5. Move instrumentation into production hot function (lowest level)

Add temporary counters around internal branches/subcalls in the suspected function.
Confirm the exact low-level operation causing native overhead.

6. Add dedicated reproducible performance test

Create a small deterministic perf test in relevant test class.
Keep correctness assertions plus one timing output line.
Ensure it exercises the exact hot path repeatedly.

7. Implement minimal expect/actual optimization

Keep common API identical.
Introduce expect in common only where needed.
Add minimal actual implementations:
- JVM path: preserve existing behavior (baseline path).
- Native path: optimized implementation for confirmed bottleneck.
Prefer smallest patch that changes only the hot operation.

Read the full file on GitHub · 111 lines

Changes

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.

16d ago First seen · 111 lines · 32 tokens per session scan A 8863a600d5dc

Subscribe to this mod's changes

speedup_native is a skill published in the GitHub repository nehemiaharchives/lucene-kmp (11 stars, last pushed 13d ago), licensed Apache-2.0. It adds 32 tokens to every session and 1,002 once invoked, about $0.0001 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 2026-10-21.

Debugging android-library apachelucene kmp kmp-library kotlin kotlin-library

speedup_native next to 6 skills from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
speedup_nativenehemiaharchives/lucene-kmp	Skill	11	13d ago	Apache-2.0	A	32	Claude Code, Codex
kotlin-tooling-native-build-performanceKotlin/kotlin-agent-skills	Skill	1.1k	7d ago	Apache-2.0	A	96	Claude Code, Codex
compose-multiplatformericrisco/rsc-harness	Skill	503	2d ago	MIT	A	80	Claude Code, Codex
compose-performancechrisbanes/skills	Skill	1.1k	yesterday	Apache-2.0	A	42	Claude Code
extract-apkniki914/zafiro	Skill	340	4d ago	MIT	A	105	Claude Code, Codex
test-hang-watchdogniki914/zafiro	Skill	340	4d ago	MIT	A	66	Claude Code, Codex
assistant-add-supportniki914/zafiro	Skill	340	4d ago	MIT	A	61	Claude Code, Codex

lucene-kmp: Skill for Claude Code