research-lit

Pass

Audited by Gen Agent Trust Hub on Nov 12, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill frequently executes local Python scripts (e.g., arxiv_fetch.py, semantic_scholar_fetch.py, deepxiv_fetch.py, exa_search.py, openalex_fetch.py, verify_papers.py, and research_wiki.py) to perform academic searches, verify data, and update a research wiki. It also uses shell commands like Glob, Grep, and Read to interact with the local file system.
  • [EXTERNAL_DOWNLOADS]: When the ARXIV_DOWNLOAD parameter is enabled, the skill uses a helper script to download PDF files from arXiv.org into the local papers/ or literature/ directories.
  • [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and analyze untrusted data from the web and research papers. This creates a surface for indirect prompt injection, where a malicious document could contain instructions designed to manipulate the agent's analysis or output. The skill attempts to mitigate this risk through a structured 'fan-out' pattern using restricted sub-agents (aris-fanout-leaf) and mandatory verification steps.
  • [DYNAMIC_EXECUTION]: The skill uses an inline Python script heredoc (`python3
  • <<'PY') to dynamically generate a fallback JSON file if the primary verification tool is unavailable. Additionally, it dynamically partitions and dispatches analysis tasks to an external agent tool (Agent(aris-fanout-leaf)`).
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 12, 2026, 06:08 AM