Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/sabbour/agentweavernpx agentmods add skills/sabbour/agentweaver/ui-harnessWrote 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/sabbour/agentweaver/ui-harness)<a href="https://agentmods.dev/skills/sabbour/agentweaver/ui-harness"><img src="https://agentmods.dev/badge/skills/sabbour/agentweaver/ui-harness/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/sabbour/agentweaver/ui-harness"><img src="https://agentmods.dev/badge/skills/sabbour/agentweaver/ui-harness.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 | $0.00043 | $0.02976 |
| Opus 5.5 | $0.00017 | $0.01190 |
| Sonnet 5.5 | $0.00009 | $0.00595 |
| Haiku 4.5 | $0.00004 | $0.00298 |
Grade A, and why
ui-harness-cli 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 20d 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 — 229 lines — stays where its author put it; the contents beside it link to each section on GitHub.
UI harness CLI contract
Use this harness to capture deterministic browser evidence for a persona against an
Agentweaver deployment. It is a driver, not a UX judge: finish emits driver P0 facts
and normalized evidence for scripts/harness-judge/; do not call subjective UX quality
verified solely from its output.
Prerequisites
From the repository root, install the harness dependencies and run its fixture tests:
node scripts/ci/shared-deps.mjs ensure --project scripts/ui-harness
npm --prefix scripts/ui-harness test
Targets are host-agnostic. Require an absolute HTTPS URL except for loopback HTTP, keep normal TLS validation enabled, and reject URL credentials/fragments. Automated browser requests and navigation stay on the configured app origin; only the explicit headful login flow may visit configured identity-provider origins.
Authentication
Agentweaver staging uses Microsoft Entra Conditional Access, which blocks plain Chromium (and device-code flow). Authentication must use the managed Chrome Default profile on Windows (enrolled device).
Option A — disposable Default-profile clone (preferred)
Close all Chrome windows first (save any open work — they will not be closed by the harness), then:
node scripts/ui-harness/login-chrome-default.mjs --base-url https://<host>.staging.<domain>
The command checks for active Chrome processes before copying anything. If Chrome still
owns the Default-profile lock, it exits with instructions to close Chrome; it does not
open an about:blank tab or fall back to the live profile. It then clones only the
managed Default profile into scripts/ui-harness/.auth/chrome-default-automation,
starts Chrome from that disposable, git-ignored clone, and deletes the clone when done.
The script navigates to --base-url before it checks or clicks Agentweaver's Sign in
with Microsoft Entra ID button. Entra SSO often completes automatically. The command
then waits for the authenticated app and saves the session without a human. If an Entra
page appears and asks for input, complete it privately in the Chrome window. The command
continues on its own after the app returns. Add --manual to wait in the Playwright
Inspector instead, and press Resume when the authenticated app is visible. The harness
never drives account selection, credentials, MFA, or consent.
What ships with it
37 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.
- .gitignore 94 B
- agent-driver-ui/session-worker.mjs 12 KB runs code
- agent-driver-ui/tools.mjs 12 KB runs code
- lib/auth.mjs 3.9 KB runs code
- lib/browser.mjs 7.9 KB runs code
- lib/chrome-default-profile.mjs 8.5 KB runs code
- lib/crossref.mjs 506 B runs code
- lib/drag.mjs 3.5 KB runs code
- lib/evidence.mjs 15 KB runs code
- lib/readiness.mjs 3.4 KB runs code
- lib/reporter-ui.mjs 2.4 KB runs code
- lib/session-runtime.mjs 17 KB runs code
- lib/session-store.mjs 1.2 KB runs code
- lib/ui-actions.mjs 7.6 KB runs code
- login-capture-chrome.mjs 416 B runs code
- login-chrome-default.mjs 4.5 KB runs code
- package-lock.json 2.3 KB
- package.json 459 B
- playwright.config.ts 199 B runs code
- README.md 6.5 KB
- test/auth.test.mjs 3.9 KB runs code
- test/browser.test.mjs 13 KB runs code
- test/chrome-default-profile.test.mjs 5.1 KB runs code
- test/drag.test.mjs 3.5 KB runs code
- test/evidence.test.mjs 2.0 KB runs code
- test/fixtures/app-shell-dom.json 256 B
- test/fixtures/auth-loading-dom.json 137 B
- test/fixtures/cluster-topology-dom.json 612 B
- test/fixtures/session-worker-close-failure.mjs 1.0 KB runs code
- test/fixtures/sign-in-dom.json 140 B
- test/init.test.mjs 461 B runs code
- test/multi-process-session.test.mjs 26 KB runs code
- test/readiness.test.mjs 5.0 KB runs code
- test/reporter-ui.test.mjs 12 KB runs code
- test/responsive-viewport.test.mjs 19 KB runs code
- test/ui-actions.test.mjs 1.9 KB runs code
- test/ui-adapter.test.mjs 4.7 KB runs code
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.
- 20d ago Changed · +23 lines 288c05cf8b08
- 27d ago Changed · +1 lines 12bc03ef6597
- 28d ago Changed · +17 lines a706204a812d
- 1mo ago Changed · +8 lines a45a2fb08c92
- 1mo ago Changed · +11 lines 39550c5f94f4
- 1mo ago Changed 293baf2fe425
- 1mo ago Changed · +2 lines 911630ff0046
- 1mo ago First seen · 167 lines · 43 tokens per session scan A f1d19393a2cc
ui-harness-cli is a skill published in the GitHub repository sabbour/agentweaver (7 stars, last pushed yesterday), licensed MIT. It adds 43 tokens to every session and 2,976 once invoked, about $0.0002 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 2026-09-03.
Other skills, from other repositories
webapp-testing
Start/reuse a local app, wait for readiness, inspect rendered state/console/network, act from observed selectors, and verify with evidence.
qa
Browser-based QA verification. Launches a real browser, navigates the app, clicks buttons, fills forms, and tests user flows. Works as a standalone skill or as a phase end condition in campaigns. Requires Playwright (optional dependency, graceful skip if not installed).
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…
run-dashboard-e2e-local-changes
Run Playwright E2E (tests/e2e/) against the docker/ all-in-one harness so it reflects LOCAL code changes, not a stale cached image.
chrome-automation
A browser-automation toolkit for controlling Chrome to visit pages, interact with forms, collect data, run tests, and create screenshots or PDFs.
browser-check
Drive a real browser and come back with a MEASUREMENT rather than an impression - console errors and >=400 responses as counts, computed styles as JSON when appearance is disputed. Use when work touches UI, when a review must verify one, or when fidelity-gate asks for its measurement.