credo-read-then-forget
Pass
Audited by Gen Agent Trust Hub on Sep 10, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process external, untrusted content such as research papers and PDFs. This creates a surface for indirect prompt injection where an attacker could embed instructions within a document to manipulate the agent's behavior.
- Ingestion points: Processes user-provided or externally sourced papers, preprints, and PDFs as specified in the outcome and scope sections.
- Boundary markers: The skill does not define specific delimiters or instructions to treat the ingested document text strictly as data, which may lead the agent to follow instructions hidden within the paper.
- Capability inventory: The skill is restricted to text analysis and summarization; no shell execution, file-writing, or network tools are requested or defined in the skill metadata.
- Sanitization: There is no evidence of sanitization or filtering logic to detect or neutralize malicious text patterns within the documents being read.
- [NO_CODE]: The skill consists exclusively of natural language prompts and instructions. It does not include scripts, executables, or configurations that would lead to code execution or environment modification.
Audit Metadata