Prevent PII leakage in PR descriptions and add structured formatting

[galel12]

Problem

The bot's generated PR descriptions expose internal PII (Red Hat email addresses, internal Jira URLs) on public GitHub repositories, and the unstructured raw dump of Jira ticket content makes PRs difficult to review.

Root Cause

The PR body was constructed by directly embedding ticket.Fields.Description, ticket.Fields.Assignee.EmailAddress, and config.Jira.BaseURL into a fmt.Sprintf call with no sanitization or formatting. There was no scrubbing layer and no structured output — the AI-generated code changes were committed, but the PR description was always a static template filled with raw Jira data.

Solution

• Updated all three AI prompt templates (claude_prompt.tmpl, gemini_prompt.tmpl, ticket_prompt.tmpl) to instruct the AI to output a structured ## PR Description section with Problem / Root Cause / Solution subsections.
• Added parsePRDescription() to extract that section from the AI's response and use it as the PR body.
• Added scrubPII() to strip email addresses and internal URLs before anything reaches GitHub.
• Removed direct embedding of assignee emails and internal Jira links from the PR body.
• Added unit tests for parsing, scrubbing, and the end-to-end PR body construction.

[Copilot]

Copilot encountered an error and was unable to review this pull request. You can try again by re-requesting a review.