run-workflow

HIGH W007: Insecure credential handling detected in skill instructions.

• Insecure credential handling detected (high risk: 1.00). The skill explicitly requires copying and presenting presigned/output URLs verbatim (HMAC-signed Azure SAS/AWS S3 URLs), which forces the LLM to handle and output secret token-bearing values directly.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). The required workflow can pull outsider-authored free text into context via Step 13–22 of references/alert-rca.md: it calls slack_get_channel_history, filters matching Slack messages (“Error Breakdown” / “Sample Workflows” payloads), and then uses that message text/fields for subsequent tool calls and synthesis—so the LLM is exposed to arbitrary Slack message body content authored by others at runtime.