asu
Pass
Audited by Gen Agent Trust Hub on Aug 31, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests and transforms untrusted user documents, such as resumes, project descriptions, and GitHub pull request lists, which could contain malicious instructions designed to influence the agent's behavior during resume generation or HR communication.
- Ingestion points: Processes existing resumes, project descriptions, and PR lists as specified in
SKILL.md. - Boundary markers: Implements a 'Claim-Evidence Ledger' with verification statuses (confirmed, to-be-confirmed) in
references/claim-evidence-ledger.mdto audit and validate user claims. - Capability inventory: Triggers transitions to downstream skills (
/make-resume,/offer) and reads/updates a shared JSON fact ledger. - Sanitization: The
references/claim-evidence-ledger.mdfile contains explicit constraints forbidding the storage of passwords, verification codes, or internal sensitive data in the ledger.
Audit Metadata