ssrn-econpapers

Pass

Audited by Gen Agent Trust Hub on Apr 6, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses curl to query academic APIs and python3 to parse the resulting data.
  • [EXTERNAL_DOWNLOADS]: Fetches research paper metadata from well-known academic services including CrossRef, SSRN, RePEc/IDEAS, and Semantic Scholar.
  • [REMOTE_CODE_EXECUTION]: Employs a piped execution pattern (curl | python3) to process API responses. The Python scripts are statically defined within the skill to extract titles, authors, and DOIs from the fetched content.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests and processes untrusted data from external research databases.
  • Ingestion points: Research paper metadata is fetched from external APIs in SKILL.md.
  • Boundary markers: No specific delimiters or warnings are used to separate fetched content from instructions.
  • Capability inventory: The skill has access to curl and python3 via the platform configuration.
  • Sanitization: Data is structurally parsed using json.load() and regular expressions, which limits the data to specific fields like titles and authors.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 6, 2026, 11:31 PM
Security Audit — agent-trust-hub — ssrn-econpapers