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 skills add mthines/agent-skills --skill jev-assertgit clone --depth 1 https://github.com/mthines/agent-skillsWrote 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/mthines/agent-skills/jev-assert)<a href="https://agentmods.dev/skills/mthines/agent-skills/jev-assert"><img src="https://agentmods.dev/badge/skills/mthines/agent-skills/jev-assert/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/mthines/agent-skills/jev-assert"><img src="https://agentmods.dev/badge/skills/mthines/agent-skills/jev-assert.svg" alt="Reviewed on agentmods" width="80" 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 | $0.00219 | $0.02240 |
| Opus 5.5 | $0.00088 | $0.00896 |
| Sonnet 5.5 | $0.00044 | $0.00448 |
| Haiku 4.5 | $0.00022 | $0.00224 |
Grade A, and why
jev-assert 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 27d 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 — 191 lines — stays where its author put it; the contents beside it link to each section on GitHub.
jev-assert
Turns a natural-language, user-observable UI expectation into an executed
semantic verdict.
It reads a page's captured text state, asks TypeSafe's Jev model one
narrow judgment, and emits a verify-behavior receipt the rest of this repo
already consumes.
It never drives a browser, never applies a fix, and never writes code — it is a
read-only verification primitive, the semantic sibling of
verify-behavior.
This
SKILL.mdis a thin index. Detailed rules live inrules/*.mdand load on demand.
What it returns
The terminal deliverable is one receipt line, byte-compatible with the
canonical vocabulary in
verification-receipt.md:
[receipt] tier: 3 | tool: jev | target: <page/spec>
[receipt] question: Does the page state show that <expectation>?
[receipt] jev: noul=0.94
[receipt] verdict: confirms
The five verdicts and when each is emitted are owned by
rules/receipt-mapping.md.
null (ran, no support) and unobtainable (could not run) are distinct and
never collapsed — the same rule verification-receipt.md enforces.
Why Jev, and why text-only
Jev is a TypeSafe System One model: it takes structured natural-language
state and returns a typed answer with a probability (Noul / Choice /
Score), not generated prose.
That is exactly the shape a test assertion needs — a decision code can gate on,
with calibrated confidence.
Jev evaluates text, not images, so the browser side must feed it an
accessibility snapshot or page text; a screenshot is never a valid input.
See rules/state-extraction.md.
Workflow
Run these three steps in order; each has a gate.
| Step | Name | Rule file | Gate |
|---|---|---|---|
| 0 | Provenance guard | rules/provenance.md |
Expectation is a user-observable OUTCOME, not a restatement of the state text |
| 1 | State extraction | rules/state-extraction.md |
Page TEXT state captured (a11y tree / page text); no screenshot |
| 2 | Jev call + mapping | rules/receipt-mapping.md |
Jev primitive resolved and mapped to exactly one receipt verdict |
What ships with it
6 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.
- 27d ago First seen · 191 lines · 219 tokens per session scan A 19dee2e58617
jev-assert is a skill published in the GitHub repository mthines/agent-skills (19 stars, last pushed today), licensed MIT. It adds 219 tokens to every session and 2,240 once invoked, about $0.0009 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 2026-09-22.
Other skills, from other repositories
browser-testing-with-devtools
Use Chrome DevTools MCP to inspect, debug, and verify anything that runs in a browser — DOM structure, console errors, network requests, performance traces, and accessibility. Use when building or fixing browser UI, diagnosing a runtime bug that isn't visible from source code alone, or verifying a fix actually works…
browser-testing-with-devtools
Tests in real browsers via Chrome DevTools MCP. Use when building or debugging anything that runs in a browser. Use when you need to inspect the DOM, capture console errors, analyze network requests, profile performance, or verify visual output with real runtime data. Requires the chrome-devtools MCP server to be…
web-test
A browser-automation tool for testing the 1C:Enterprise web client. It uses Playwright to navigate sections, fill forms, read tables and reports, and perform user actions in a browser.
actionbook-web-test
Run browser-based web tests against websites using Actionbook CLI. Activate when the user wants to test a website workflow, run smoke tests, verify a user flow, check if a web application works, run regression tests, or validate browser-based interactions. Supports test definition, execution, assertion, reporting, and…
playwright-visual-testing
Add, repair, or review Playwright visual regression tests for browser-facing .NET apps, including screenshot baselines, Pixelmatch thresholds, deterministic rendering, and GitHub Actions artifacts. USE FOR: toHaveScreenshot, page.screenshot visual checks, Pixelmatch/pngjs comparison scripts, visual baseline updates…
dogfood
This skill guides you through systematic exploratory QA testing of web applications using the browser toolset. You will navigate the application, interact with elements, capture evidence of issues, and produce a structured bug report.