bug-screenshot-annotation

Pass

Audited by Gen Agent Trust Hub on Sep 17, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill utilizes python3 for image cropping and serving, playwright-cli for screen capture, and bun for Jira integration. These operations are performed locally and support the primary function of evidence generation.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests external screenshots which may contain adversarial text, presenting a vulnerability surface for indirect prompt injection.
  • Ingestion points: Raw bug screenshots identified in SKILL.md (Process Step 1).
  • Boundary markers: No delimiters or instructions to ignore embedded content are implemented in the rendering workflow.
  • Capability inventory: Local file access via PIL, loopback HTTP serving, browser automation via Playwright, and Jira API integration via bun as documented in SKILL.md.
  • Sanitization: There is no documented validation or sanitization for text content placed into HTML callout boxes.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 17, 2026, 08:14 PM
Security Audit — agent-trust-hub — bug-screenshot-annotation