azure-functions-best-practices
Pass
Audited by Gen Agent Trust Hub on Aug 7, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [SAFE]: The skill implements a secure-by-default posture by requiring the redaction of connection strings, SAS tokens, and keys in all outputs and reports.
- [SAFE]: Human-in-the-loop controls are strictly enforced, with mandatory user approval required before the agent can modify networking, identity, or application settings.
- [SAFE]: The skill employs least-privilege data collection by specifically querying for app setting names while excluding their sensitive values from the retrieval process.
- [PROMPT_INJECTION]: The skill processes external configuration data, creating an attack surface for indirect prompt injection. However, this is mitigated by a structured reporting workflow and explicit human review requirements.
- Ingestion points: Azure CLI tool outputs (az functionapp) and the get_azure_bestpractices MCP tool.
- Boundary markers: Use of a report-first workflow and explicit user approval gates for all actions.
- Capability inventory: Utilization of Azure CLI tools for resource inspection and generation of Bicep/Terraform remediation patches.
- Sanitization: All potentially sensitive data is filtered or redacted as per the skill's core principles.
Audit Metadata