skills/yashas-30/nyx/search-first/Gen Agent Trust Hub

search-first

Pass

Audited by Gen Agent Trust Hub on Jun 24, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill instructs the agent to perform research by searching external sources such as npm, PyPI, and GitHub. This creates a surface where an attacker could potentially influence the agent's actions through malicious package descriptions or repository content.\n
  • Ingestion points: Search results from npm, PyPI, GitHub, and the general web (File: SKILL.md, Step 2: Parallel Search).\n
  • Boundary markers: No explicit instructions or markers are provided to the agent to treat external search results as untrusted content or to ignore embedded instructions.\n
  • Capability inventory: The agent is empowered to execute shell commands, specifically package managers (npm install, pip install), file system operations (ls, rg), and the GitHub CLI (gh).\n
  • Sanitization: There is no mention of validating or sanitizing the content retrieved from external searches before evaluation or implementation steps.\n- [COMMAND_EXECUTION]: The skill guides the agent to use local CLI tools like 'rg', 'ls', 'npm', 'python -m pip', and 'gh' to verify the development environment and search for existing solutions during the Tool Availability Preflight and Quick Mode steps.\n- [EXTERNAL_DOWNLOADS]: The skill advocates for finding and installing external dependencies to solve coding problems. It references numerous well-known and reputable packages such as eslint, prettier, httpx, and zod as examples of high-quality candidates for adoption.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 24, 2026, 08:21 AM
Security Audit — agent-trust-hub — search-first