The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from the Juicebox API, creating a potential surface for indirect prompt injection if external profile records contain malicious instructions.
Ingestion points: importPeopleDataset (SKILL.md) fetches profile records from api.juicebox.ai.
Boundary markers: Present; the skill instructs to inventory fields, quarantine unknown authorities, and run suppression validation on synthetic batches.
Capability inventory: The skill is limited to HTTP requests and local data processing; no dangerous shell command execution or dynamic code evaluation capabilities were identified.
Sanitization: Present; includes schema validation in validateProfile and PII sanitization logic in exportAnalysisResults.

juicebox-data-handling