delete-article
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest external data which is then used to modify repository configuration files.
- Ingestion points: The agent accepts redirect URLs from user input or search results generated by the
microsoft-learn-groundingtool (referenced in SKILL.md). - Boundary markers: The instructions require the agent to propose selected redirect targets to the user for explicit confirmation before any repository edits occur.
- Capability inventory: The skill performs file deletions (Markdown, YAML, and images), repository-wide string searches, and modifications to sensitive configuration files including
.openpublishing.redirection.json,docs/toc.yml, anddocs/index.yml. - Sanitization: The workflow relies on mandatory human validation of external targets to ensure the integrity of the documentation and redirection logic.
Audit Metadata