Open-Inspect is a hosted system that lets coding agents work on software tasks in background development environments. Teams use it to connect agents with repositories, communication tools, scheduled events, and parallel sandboxes. The catalogue contains add-ons associated with this system.
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 ColeMurray/background-agents --skill visual-verificationgit clone --depth 1 https://github.com/ColeMurray/background-agentsWrote 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/colemurray/background-agents/visual-verification)<a href="https://agentmods.dev/skills/colemurray/background-agents/visual-verification"><img src="https://agentmods.dev/badge/skills/colemurray/background-agents/visual-verification/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/colemurray/background-agents/visual-verification"><img src="https://agentmods.dev/badge/skills/colemurray/background-agents/visual-verification.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00014 | $0.01783 |
| Opus 5.5 | $0.00006 | $0.00713 |
| Sonnet 5.5 | $0.00003 | $0.00357 |
| Haiku 4.5 | $0.00001 | $0.00178 |
Grade A, and why
visual-verification 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 1mo 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 — 208 lines — stays where its author put it; the contents beside it link to each section on GitHub.
visual-verification
Use this skill when the goal is to verify UI changes inside the application and return visual evidence in the Open-Inspect session.
agent-browser remains the low-level browser tool. This skill defines the workflow contract for
using it reliably.
Key Fact
upload-media is a bash command installed on PATH. Run it with your Bash tool, not as an MCP
tool or tool binding. For videos, use agent-browser record directly, then probe and upload the
resulting MP4 with upload-media.
When To Use It
- Verify a UI change after editing code
- Capture before/after screenshots for comparison
- Confirm responsive layout differences at a chosen viewport
- Produce an uploaded screenshot or short video artifact the user can review in-session
Success Criteria
The task is not complete until all of these are true:
- The changed UI is opened in the browser.
- The capture mode is chosen explicitly: viewport screenshot, full-page screenshot, or video.
- The viewport is set explicitly or reported as a deliberate default.
- A screenshot or video is uploaded in the same prompt.
- The returned
artifactIdis reported back to the user. - The response states what was verified and what dimensions/mode were used.
Required Workflow
- Open the target page with
agent-browser open. - If viewport matters, set it explicitly with
agent-browser set viewport <width> <height>. - Wait for the page to settle before capture.
- Choose one of:
- Viewport screenshot for above-the-fold or device-specific review
- Full-page screenshot for full document review
- Video recording for interaction flows, animations, transitions, or multi-step behavior
- Upload the capture immediately with matching metadata.
- Report the result with the artifact ID and actual capture settings.
Default Decision Rules
- Use a viewport screenshot when validating a specific visible state, modal, interaction, or desktop/mobile layout.
- Use a full-page screenshot when the user asks for the whole page or when vertical content is part of the verification.
- If the user names a device or screen size, set the viewport explicitly.
- If the user does not specify dimensions and layout matters, choose a reasonable viewport and report it.
- If the screenshot is intended to prove a fix, prefer stating exactly what was checked, not only that a screenshot was taken.
- Use a video when the proof depends on seeing interaction over time, such as opening a menu, dragging, typing, navigating between states, or watching an animation complete.
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.
- 1mo ago First seen · 208 lines · 14 tokens per session scan A 0b47d292f463
visual-verification is a skill published in the GitHub repository ColeMurray/background-agents (3,375 stars, last pushed today), licensed MIT. It adds 14 tokens to every session and 1,783 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 2026-08-30.
Other skills, from other repositories
qa
Use when you must verify a change actually works through the REAL browser UI — acceptance, end-to-end, smoke tests, form interactions, navigation, login flows — driven by the agent-browser tool. Invoke when "click through it", "does the UI work", "smoke test the page", or browser-level evidence (screenshots…
verify-omnigent
Drive Omnigent the way a user does and prove a behavior with recorded evidence, using an isolated server, runner, and mock model plus a feature map of every user entry point. Load before reproducing a user-facing bug, before claiming a fix works, or when reviewing whether a change covered every surface (session vs.…
test-master
Generates test files, creates mocking strategies, analyzes code coverage, designs test architectures, and produces test plans and defect reports across functional, performance, and security testing disciplines. Use when writing unit tests, integration tests, or E2E tests; creating test strategies or automation…
dogfood
Systematically explore and test a mobile app on iOS/Android with agent-device to find bugs, UX issues, and other problems. Use when asked to dogfood, QA, exploratory test, find issues, bug hunt, or test this app on mobile.
cua-driver
Use Cua Driver for desktop or browser tasks that are awkward or unavailable through Bash/APIs, or when the user explicitly wants GUI interaction: app testing, visual bug reproduction, form filling, calendar entry, screenshots, and demo recording. Also covers Cua setup; not OpenAI Codex Computer Use or web research.
critique
Adversarial design critique of the current work — render it, look at it, and argue for rejection. Run after the gates are green, never instead of them.