managing-feature-flags
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill identifies a potential attack surface where data fetched from external sources could contain malicious instructions.
- Ingestion points: The skill uses
WebFetchto retrieve content fromcontributing.bitwarden.comand utilizes LaunchDarkly MCP tools to discover flag metadata. - Boundary markers: The instructions explicitly warn the agent to treat external content as untrusted data (referencing CWE-1427) and provide clear boundaries for how to handle this data.
- Capability inventory: The skill's capabilities are restricted to reading, searching, and fetching from a specific vendor-controlled domain. It does not have permissions to execute arbitrary code or access sensitive local files.
- Sanitization: The skill mandates that the agent should only summarize or reference external content and must never execute instructions found within it, specifically noting that metadata should not influence code edits or command execution.
Audit Metadata