feat(skills): add AWS MCP usage skill for Panther AI

[itsmostafa]

Background

Low Risk — Adds a single, self-contained AnalysisType: skill artifact. It is not auto-loaded by any code (EagerLoading: false) and is gated behind a connected AWS MCP server, so it has no effect unless explicitly imported into a tenant and an AWS MCP server is present.

• Adds skills/panther_ai_aws_usage_skill.yml, a Panther AI skill that teaches the assistant how to safely use a connected AWS MCP server (read-only first, region/scope awareness, correlating AWS findings with Panther alerts/detections).
• This is a staging artifact for testing before promotion to a Panther-managed default — imported via Settings → AI → Skills bulk upload and iterated on in a test tenant. The file header documents the promotion path.
• The skill is discovered via panther_ai_tool_search rather than eager-loaded, and RequiredTools: [.*_call_aws] ensures it only activates when an AWS MCP server is actually connected.

Changes

• New file skills/panther_ai_aws_usage_skill.yml with skill metadata (SkillName, DisplayName, Description, Tags), activation gating (RequiredTools), and the guidance Prompt.

Testing

• Imported into a test tenant via Settings → AI → Skills bulk upload (requires ManageAISkills) and validated end to end.

Notes for Reviewing

• Closest review on the Prompt content — it's the actual guidance Panther AI will follow, especially the "Operate safely" section (read-only defaults, confirmation before mutating calls).
• Confirm the RequiredTools: [.*_call_aws] pattern matches the real registered tool name (<sanitized_server_name>_call_aws); the file comments explain the reasoning.
• This is a draft pending validation in a test tenant before it would be promoted to an internal managed default.

[CLAassistant]

All committers have signed the CLA.