dx-references
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill exclusively reads documentation from a scoped internal directory (
${CLAUDE_SKILL_DIR}/references/). It does not engage in network activity, external downloads, or execution of arbitrary code. - [SAFE]: The skill package includes a dedicated security policy file (
references/untrusted-content.md) which proactively instructs the agent to treat data from external searches or cited research as non-executable text, specifically warning against following embedded instructions or commands. - [SAFE]: Operational guidelines provided in the references (e.g., in
start-check.md) follow best practices for developer tools, such as requiring user confirmation before committing to default branches and respecting user-defined naming conventions.
Audit Metadata