sending-emails-with-outlook

Warn

Audited by Gen Agent Trust Hub on Nov 16, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill provides a bash command template in SKILL.md that directly interpolates a user-controlled <task> variable into a shell string (claude -p "<task>"). This pattern is susceptible to shell command injection if the task input contains shell metacharacters such as semicolons, pipes, or backticks.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data from email drafts and search results, creating an attack surface for indirect prompt injection.
  • Ingestion points: The skill retrieves existing email content and drafts using the outlook_email_search tool and draft management functions in SKILL.md.
  • Boundary markers: The instructions do not specify the use of delimiters (like XML tags or triple quotes) or warnings to ignore instructions embedded within the ingested email content.
  • Capability inventory: The skill has the capability to create, update, and send emails via the outlook_create_draft, outlook_update_draft, and outlook_send_mail tools mentioned in SKILL.md.
  • Sanitization: No sanitization is performed on the plain text or HTML content of retrieved emails to neutralize potential malicious instructions, although some HTML attribute stripping is mentioned for technical compatibility.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Nov 16, 2026, 05:24 PM