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 agentmods add skills/ngocsangyem/meowkit/benchmarknpx skills add ngocsangyem/MeowKit --skill benchmarkgit clone --depth 1 https://github.com/ngocsangyem/MeowKitWrote 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/ngocsangyem/meowkit/benchmark)<a href="https://agentmods.dev/skills/ngocsangyem/meowkit/benchmark"><img src="https://agentmods.dev/badge/skills/ngocsangyem/meowkit/benchmark.svg" alt="Measured on agentmods" height="20"></a>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 | $0.00026 | $0.02673 |
| Opus 5 | $0.00013 | $0.01337 |
| Sonnet 5 | $0.00005 | $0.00535 |
| Haiku 4.5 | $0.00003 | $0.00267 |
Grade A, and why
mk:benchmark 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 2d 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 — 222 lines — stays where its author put it; the contents beside it link to each section on GitHub.
mk:benchmark — Experimental Harness Canary Suite
Measures harness performance against a small set of ground-truth tasks. Provides the empirical signal that the dead-weight audit (per .claude/rules/dead-weight-audit-rules.md) consumes to make load-bearing decisions about each harness component.
When to Use
Activate when:
- User runs
/mk:benchmark run(default = quick tier, 5 tasks, ≤$5) - User runs
/mk:benchmark run --full(quick tier + 1 heavy task, ≤$30) - User runs
/mk:benchmark compare <run-id-a> <run-id-b>(delta table) - Before applying a harness change (baseline)
- After applying a harness change (verify delta)
- During the dead-weight audit playbook (component enable/disable cycles)
Skip when:
- The harness has been run end-to-end manually within the last hour (use that data instead)
- Budget cap is hit before the suite finishes (record partial result, alert)
Hard Constraints
- Quick tier ≤$5 total cost. Hard block if projected cost exceeds.
- Full tier ≤$30 total cost. Hard block if projected cost exceeds.
--fullis opt-in. The heavy task (06-small-app-build) requires explicit--fullflag because it triggersmk:autobuildwhich can run for hours. Refuses to run without the flag.- NOT a replacement for unit tests. This is harness-level measurement only.
- Results recorded in trace-log.jsonl as
event=benchmark_resultrecords, tagged withbenchmark_version+harness_version+model_version.
Subcommands
| Subcommand | Purpose | Tier | Cost cap |
|---|---|---|---|
run |
Execute the quick tier (5 tasks) and record scores | quick | $5 |
run --full |
Execute quick + heavy tier (6 tasks total) | full | $30 |
compare <a> <b> |
Show per-task delta between two prior runs | — | (free, reads cache) |
Tier Layout
.claude/benchmarks/
├── README.md ← how to use + add tasks
├── canary/
│ ├── quick/ ← default tier (5 tasks, ≤$5)
│ │ ├── 01-react-component-spec.md
│ │ ├── 02-api-endpoint-spec.md
│ │ ├── 03-bug-fix-spec.md
│ │ ├── 04-refactor-spec.md
│ │ └── 05-tdd-feature-spec.md
│ └── full/ ← --full only (1 task, ~$25)
│ └── 06-small-app-build-spec.md
└── results/ ← per-run JSON dumps
What ships with it
3 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 2d ago First seen · 222 lines · 26 tokens per session scan A 7e76ef692ba0
mk:benchmark is a skill published in the GitHub repository ngocsangyem/MeowKit (14 stars, last pushed 1mo ago), licensed MIT. It adds 26 tokens to every session and 2,673 once invoked, about $0.0001 per session on Opus 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 2026-09-03.
Other skills, from other repositories
benchmarking
Skill "benchmarking" from chaterm/terminal-skills, covering 性能基准测试, 概述, sysbench, 安装 and debian/ubuntu.
http-load-tester
HTTP 阶梯式并发压测工具，自动执行多级并发压力测试，采集 p50/p90/p99 延迟百分位数，分析性能拐点（如延迟激增、吞吐饱和或错误率飙升），并输出结构化报告与最优并发建议。使用纯 Python 标准库，零外部依赖。当用户提及 HTTP 压测、基准测试、ab、wrk、并发测试、阶梯压测、延迟百分位、p99、拐点分析、性能测试、QPS、RPS 或容量规划等关键词或场景时触发。.
http-load-profiler
Run stepped HTTP load tests with ab/wrk, ramping concurrency levels to collect p50/p90/p99 latency, detect performance inflection points, and recommend optimal concurrency. Triggered by requests like 'load test this URL', 'benchmark my API', 'find the max concurrency', or mentions of p99 latency, throughput…
exploratory-data-analysis
Perform bounded, local exploratory analysis of explicitly supported scientific files. Use for redacted CSV/TSV/JSON profiles; optional NumPy, HDF5, FASTA/FASTQ, and basic image metadata inspection; missingness/leakage audits; outlier and transformation sensitivity; and rigorous EDA report scaffolds. Other domain…
golden-rss
Use when testing the rss golden build.
google-ads-audit
Google Ads account audit and business context setup. Run this first — it gathers business information, analyzes account health, and saves context that all other ads skills reuse. Trigger on "audit my ads", "ads audit", "set up my ads", "onboard", "account overview", "how's my account", "ads health check", "what should…