skills/beclab/olares/olares-search/Gen Agent Trust Hub

olares-search

Pass

Audited by Gen Agent Trust Hub on Jun 19, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests and processes snippets of content from user-indexed files without explicit boundary markers or sanitization. If a file contains malicious instructions, the agent may follow them when reviewing search results.\n
  • Ingestion points: olares-cli search output snippets (SKILL.md).\n
  • Boundary markers: Absent; the instructions do not define delimiters or protective instructions for untrusted file content.\n
  • Capability inventory: The skill belongs to a suite that includes olares-files (read/write/download) and olares-settings, which provide significant interaction surfaces that could be targeted by injected instructions.\n
  • Sanitization: Absent; content hits are passed verbatim to the agent as part of the search results.\n- [COMMAND_EXECUTION]: The skill documents the execution of the olares-cli binary to perform search operations.\n
  • Evidence: Functional description of olares-cli search used to query the search3 index. This tool is a vendor resource provided by beclab.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 19, 2026, 07:08 PM
Security Audit — agent-trust-hub — olares-search