browser-screenshot-capture
Capture browser-based evidence for AML review in a way that is reproducible and human-reviewable.
SKILL.mdskills/connectors/browser-screenshot-capture/SKILL.md
GitHubBrowser Screenshot Capture Connector Skill
Purpose
Capture browser-based evidence for AML review in a way that is reproducible and human-reviewable.
When to Use
Use when an AML workflow requires a visual record of a web page, search result, registry page, sanctions result, adverse media result, website page, domain lookup or source document.
Required Inputs
- case identifier;
- subject under review;
- source URL or page title;
- reason for capture;
- output folder;
- date/time if available.
Workflow
- Verify that the page relates to the correct subject or source query.
- Capture the visible page area required by policy.
- If the result spans multiple pages, capture each material page separately.
- Ensure the URL/address bar or source identifier is visible where technically possible.
- Save the screenshot using the implementation profile naming rule.
- Add an entry to the evidence log.
- Mark whether the capture is complete, partial or limited.
Evidence Metadata
Record:
- source name;
- URL or source identifier;
- access date;
- query/search term;
- screenshot file name;
- evidence quality label;
- limitations.
Red Flag Routing
The screenshot connector does not decide whether a finding is material. If the captured page contains a red flag under the selected AML skill or implementation profile, route the case to manual review.
Agent Must Not
- crop out material context;
- edit or enhance evidence in a way that changes meaning;
- summarize a screenshot as a final compliance decision;
- omit visible uncertainty.