db-migrate
Warn
Audited by Gen Agent Trust Hub on Apr 29, 2026
Risk Level: MEDIUMPROMPT_INJECTIONCOMMAND_EXECUTIONREMOTE_CODE_EXECUTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses repetitive, high-pressure instructional patterns like 'MANDATORY IMPORTANT MUST ATTENTION' and 'HARD-GATE' to override default agent behavior and enforce specific task-breaking workflows.
- [COMMAND_EXECUTION]: The skill takes user-provided
$ARGUMENTSand maps them to shell commands (e.g.,dotnet ef migrations add <MigrationName>). This creates a surface for command injection if input is not strictly validated before execution. - [REMOTE_CODE_EXECUTION]: The skill instructs the agent to execute a local Python script (
.claude/scripts/code_graph) that is not part of the skill's own package. This represents an unverifiable dependency that could perform arbitrary operations. - [DATA_EXFILTRATION]: The skill documents an ingestion point for untrusted data in
docs/project-reference/domain-entities-reference.md, noting that content is 'auto-injected by hook'. This presents an indirect prompt injection surface.
Audit Metadata