Borrowing it
Nothing to install: this file belongs to PIsberg/vibetags. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/PIsberg/vibetags/main/.claude/skills/load-tests/SKILL.mdgit clone --depth 1 https://github.com/PIsberg/vibetagsWrote 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/pisberg/vibetags/load-tests)<a href="https://agentmods.dev/skills/pisberg/vibetags/load-tests"><img src="https://agentmods.dev/badge/skills/pisberg/vibetags/load-tests/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.
<a href="https://agentmods.dev/skills/pisberg/vibetags/load-tests"><img src="https://agentmods.dev/badge/skills/pisberg/vibetags/load-tests.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Excessive Agency · line 149 Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.Fix: Add human-in-the-loop confirmation for destructive, irreversible, or high-impact operations. Never auto-execute commands that modify files, send data, or alter system state.
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 | $0.00094 | $0.04107 |
| Opus 5.5 | $0.00038 | $0.01643 |
| Sonnet 5.5 | $0.00019 | $0.00821 |
| Haiku 4.5 | $0.00009 | $0.00411 |
Grade A, and why
load-tests 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 25d 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 — 244 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Load tests
The harness lives in load-tests/. It is a standalone Maven project that compiles synthetic
annotated sources with and without the processor and subtracts. load-tests/README.md documents
the harness; load-tests/results/README.md documents the baselines. This skill is the judgement
around both: which run answers which question, and which of the resulting numbers may be quoted.
The single most important fact about this harness: most of what it prints is noise on a developer workstation. Two runs of an identical build have differed by up to 1.93x on the JMH hot path. Only the allocation numbers are stable enough to make claims from. Everything below follows from that.
Step 1 — Decide which question is being asked
"Does this change make the processor slower?" Only the allocation sweep can answer that on an ordinary machine, and only if both versions are measured back-to-back in one sitting. See Step 4.
"Capture a baseline for the release." The full capture in Step 3, recorded under
load-tests/results/<version>/ with an env.txt that says what the machine was doing.
"Prove this specific optimisation works." Neither sweep will show it unless it is large.
Write a targeted JMH benchmark or a targeted stress test instead — WriteCacheHitBenchmark and
SignatureCaptureStressTest are the two worked examples in the repo, and both exist because the
general sweeps could not see the effect they were built to measure.
"Where did the curve move over the last few releases?" The release-trend plots cannot answer
it: they overlay per-release captures taken months apart. Sweep every release you care about in
one sitting instead, switching only -Dprocessor.version, and plot it with
tools/plot-release-history.py. Done over 1.2.5..main on 2026-09-22 this found a 6 % step at
1.3.2 that eight releases of cross-day baselines had not shown. If it finds one, bisect it to a
commit with git worktree rather than guessing from the changelog. The guess that session made
from the changelog was wrong, and the commit the measurement named was one whose javadoc asserted
the cost was negligible.
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.
- 25d ago Changed · +2 lines 21891115f9a3
- 26d ago Changed · +3 lines eea2daea434f
- 29d ago Changed · +68 lines 51994ae3341e
- 1mo ago Changed · +11 lines 472810c371a8
- 1mo ago First seen · 160 lines · 94 tokens per session scan A 347d84d41022
load-tests is a skill published in the GitHub repository PIsberg/vibetags (25 stars, last pushed today), licensed MIT. It adds 94 tokens to every session and 4,107 once invoked, about $0.0004 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 2026-08-30.
Other skills, from other repositories
agent-memory-testing
Use this skill when you need evidence-bounded memory write/read/update/delete, retention, contamination, isolation, provenance, and forgetting behavior; triggers include Agent 记忆 and Agent memory.
cgs-soak-test
Use for soak test tasks that plan longer stability checks for memory, performance drift, save/load, networking, and live-ops loops; produce verification evidence, changed or proposed files, and handoff boundaries.
tdd-with-memory
A test-first implementation workflow that recalls past decisions, conventions, and lessons before coding. TDD, or test-driven development, means writing a failing test first, then implementing the code and finally cleaning it up.
lore-onboard
First-time setup wizard for LoreConvo. Verifies the MCP server is connected, the database is accessible, hooks are configured, and runs a test save/load cycle. Use when the user says "set up loreconvo", "onboard", "/lore-onboard", "verify loreconvo", "test loreconvo setup", or after first installing the plugin.
wshobson-backtesting-frameworks
Skill "wshobson-backtesting-frameworks" from ItamarZand88/awesome-agent-conventions, covering backtesting frameworks, when to use this skill, core concepts, 1. backtesting biases and 2. proper backtest structure.
mnemon
Persistent memory CLI for LLM agents. Store facts, recall past knowledge, link related memories, manage lifecycle.