search-first

Pass

Audited by Gen Agent Trust Hub on Oct 14, 2026

Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The skill instructs the agent to read sensitive platform configuration files, specifically ~/.claude/settings.json. These files often contain API keys, authentication tokens, and environment configurations that could be exposed if the agent is further compromised.
  • [INDIRECT_PROMPT_INJECTION]: The workflow relies on processing untrusted data from external registries (npm, PyPI) and web search results to inform critical decisions like package installation.
  • Ingestion points: Search results from npm, PyPI, GitHub, and the web are ingested into the agent's context (SKILL.md).
  • Boundary markers: No specific delimiters or instructions are provided to the agent to treat external search results as untrusted data or to ignore embedded instructions within them.
  • Capability inventory: The skill allows the agent to install packages (npm install, pip install), read local configuration files, and write/modify code.
  • Sanitization: The skill lacks explicit sanitization or validation steps for the content retrieved during the research phase, creating a surface for malicious actors to influence the agent's actions through manipulated package metadata or search results.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 14, 2026, 02:58 PM