search

Pass

Audited by Gen Agent Trust Hub on Jul 26, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses dynamic context injection via the !date command to include a timestamp in the output. This is a benign use of the feature for documenting when a search was performed.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes and incorporates untrusted data from external search engines (Exa, grep.app) and third-party repositories. Malicious instructions embedded in this external content could potentially influence the agent's behavior.
  • Ingestion points: External search results from Exa and grep.app, and content from repositories cloned during deep-dive analysis (SKILL.md).
  • Boundary markers: The skill lacks explicit markers or instructions to isolate or ignore potentially malicious prompts within the ingested external data.
  • Capability inventory: The skill uses local file system tools (grep, glob, ast_grep), external network searches, and has the ability to write to the file system using obsidian_append_content.
  • Sanitization: No sanitization or validation of the retrieved external content is specified before it is processed by the agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 26, 2026, 12:10 AM
Security Audit — agent-trust-hub — search