abstract-searcher

Pass

Audited by Gen Agent Trust Hub on Mar 7, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: Fetches academic paper data and abstracts from well-known services including arXiv, Semantic Scholar, CrossRef, and OpenAlex. These are trusted academic sources and the operations are consistent with the skill's purpose.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests and processes paper abstracts and browser snapshots from external sources without sufficient delimiters. Evidence chain:\n
  • Ingestion points: API responses in scripts/add_abstracts.py and browser snapshots described in SKILL.md.\n
  • Boundary markers: Absent; the fetched content is not wrapped in protective delimiters or accompanied by instructions to ignore embedded prompts.\n
  • Capability inventory: Performs network requests via urllib.request and utilizes browser automation tools (tabs, open, snapshot, act) via a Chrome profile session.\n
  • Sanitization: The clean_abstract function in scripts/add_abstracts.py provides minimal sanitization by removing newlines and excessive whitespace, which does not prevent instruction execution.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 7, 2026, 07:29 PM