benefits-processing
Pass
Audited by Gen Agent Trust Hub on Mar 23, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill's primary function is to perform a technical and regulatory audit of a codebase related to government benefits processing (e.g., SNAP, Medicaid). It identifies components such as rules engines, integration endpoints, and workflow logic without executing external code.
- [DATA_EXPOSURE_AND_EXFILTRATION]: The skill contains explicit 'DO NOT' instructions that forbid the inclusion of real PII, SSNs, or applicant data in its output, effectively mitigating the risk of accidental data exposure during the analysis of sensitive systems.
- [PERSISTENCE_MECHANISMS]: The skill implements a 'Self-evolution Telemetry' feature that records execution metadata (outcome, iterations, bottlenecks) to a local directory (
~/.claude/projects/). This is a standard logging pattern for tracking skill performance within the CLAUDE_CODE environment and does not represent a malicious persistence or backdooring attempt. - [INDIRECT_PROMPT_INJECTION]: While the skill ingests untrusted data in the form of the codebase it analyzes, it lacks high-risk capabilities such as network access or arbitrary command execution, limiting the impact of any potential indirect injection found within the analyzed files.
Audit Metadata