[{"data":1,"prerenderedAt":4},["ShallowReactive",2],{"readme:skillspector":3},"\u003Ch1>SkillSpector\u003C\u002Fh1>\n\u003Cp>\u003Cstrong>Security scanner for AI agent skills.\u003C\u002Fstrong> Detect vulnerabilities, malicious patterns, and security risks before installing agent skills.\u003C\u002Fp>\n\u003Cp>\u003Ca href=\"https:\u002F\u002Fwww.python.org\u002Fdownloads\u002F\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002Fpython-3.12+-blue.svg\" alt=\"Python 3.12+\" \u002F>\u003C\u002Fa>\n\u003Ca href=\"https:\u002F\u002Fwww.apache.org\u002Flicenses\u002FLICENSE-2.0\" rel=\"nofollow ugc noopener\">\u003Cimg src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-Apache%202.0-blue.svg\" alt=\"License: Apache 2.0\" \u002F>\u003C\u002Fa>\u003C\u002Fp>\n\u003Ch2>Overview\u003C\u002Fh2>\n\u003Cp>AI agent skills (used by Claude Code, Codex CLI, Gemini CLI, etc.) execute with implicit trust and minimal vetting. In the 31,132-skill analyzed subset of the research dataset, \u003Cstrong>26.1% of skills contain vulnerabilities\u003C\u002Fstrong> and \u003Cstrong>5.2% show likely malicious intent\u003C\u002Fstrong>.\u003C\u002Fp>\n\u003Cp>SkillSpector helps you answer: \u003Cstrong>\"Is this skill safe to install?\"\u003C\u002Fstrong>\u003C\u002Fp>\n\u003Cp>SkillSpector is part of the \u003Ca href=\"https:\u002F\u002Fdocs.nvidia.com\u002Fskills\u002F\" rel=\"nofollow ugc noopener\">NVIDIA Verified Skills pipeline\u003C\u002Fa>, which scans, evaluates, and signs agent skills before publication. Skills that pass are published to the \u003Ca href=\"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskills\" rel=\"nofollow ugc noopener\">NVIDIA skills catalog\u003C\u002Fa>.\u003C\u002Fp>\n\u003Ch2>Documentation\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>\u003Cstrong>\u003Ca href=\"https:\u002F\u002Fdocs.nvidia.com\u002Fskills\u002Fscanning-agent-skills\" rel=\"nofollow ugc noopener\">Scan agent skills before installation\u003C\u002Fa>\u003C\u002Fstrong> — Hosted guide: when to scan, how to read a report, and how to gate installs.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskillspector\u002Fblob\u002FHEAD\u002Fdocs\u002FDEVELOPMENT.md\" rel=\"nofollow ugc noopener\">Development guide\u003C\u002Fa>\u003C\u002Fstrong> — Architecture, package layout, and how to extend the analyzer pipeline.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskillspector\u002Fblob\u002FHEAD\u002Fdocs\u002FANALYSIS_RESOURCE_BOUNDS.md\" rel=\"nofollow ugc noopener\">Analysis resource bounds\u003C\u002Fa>\u003C\u002Fstrong> — Fail-closed bundle, parser, nested-artifact, ledger, and finding ceilings.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskillspector\u002Fblob\u002FHEAD\u002Fdocs\u002FPI_EXTENSION.md\" rel=\"nofollow ugc noopener\">Pi extension\u003C\u002Fa>\u003C\u002Fstrong> — Install SkillSpector as a Pi tool for scanning skills from inside agent sessions.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>\u003Ca href=\"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskillspector\u002Fblob\u002FHEAD\u002Fdocs\u002FOPENCODE_EXTENSION.md\" rel=\"nofollow ugc noopener\">OpenCode extension\u003C\u002Fa>\u003C\u002Fstrong> — Install SkillSpector as an OpenCode tool and \u003Ccode>\u002Fskillspector\u003C\u002Fcode> command for scanning skills from inside agent sessions.\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch2>Features\u003C\u002Fh2>\n\u003Cul>\n\u003Cli>\u003Cstrong>Multi-format input\u003C\u002Fstrong>: Scan Git repos, URLs, zip files, directories, or single files\u003C\u002Fli>\n\u003Cli>\u003Cstrong>71 vulnerability patterns\u003C\u002Fstrong> across 17 categories: prompt injection, data exfiltration, privilege escalation, supply chain, excessive agency, output handling, system prompt leakage, memory poisoning, tool misuse, rogue agent, anti-refusal, trigger abuse, dangerous code (AST), taint tracking, YARA signatures, MCP least privilege, and MCP tool poisoning\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Two-stage analysis\u003C\u002Fstrong>: Fast static analysis + optional LLM semantic evaluation\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Live vulnerability lookups\u003C\u002Fstrong>: SC4 queries \u003Ca href=\"https:\u002F\u002Fosv.dev\" rel=\"nofollow ugc noopener\">OSV.dev\u003C\u002Fa> for real-time CVE data with automatic offline fallback\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Multiple output formats\u003C\u002Fstrong>: Terminal, JSON, Markdown, and SARIF reports\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Risk scoring\u003C\u002Fstrong>: 0-100 score with severity labels and clear recommendations\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Baseline \u002F false-positive suppression\u003C\u002Fstrong>: Accept known findings via a glob-rule or fingerprint baseline so re-scans surface only \u003Cem>new\u003C\u002Fem> issues (\u003Ca href=\"https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskillspector\u002Fblob\u002FHEAD\u002Fdocs\u002FSUPPRESSION.md\" rel=\"nofollow ugc noopener\">docs\u003C\u002Fa>)\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch2>Quick Start\u003C\u002Fh2>\n\u003Ch3>Installation\u003C\u002Fh3>\n\u003Cblockquote>\n\u003Cp>\u003Cstrong>Open-source software notice:\u003C\u002Fstrong> This project will download and install additional third-party open source software projects. Review the license terms of these open source projects before use.\u003C\u002Fp>\n\u003C\u002Fblockquote>\n\u003Cp>Create and activate a virtual environment first (all \u003Ccode>make\u003C\u002Fcode> targets assume the venv is active). Use \u003Cstrong>uv\u003C\u002Fstrong> or \u003Cstrong>pip\u003C\u002Fstrong>; the Makefile uses \u003Ccode>uv\u003C\u002Fcode> if available, otherwise \u003Ccode>pip\u003C\u002Fcode>.\u003C\u002Fp>\n\u003Cp>\u003Cstrong>Quick install with uv (CLI-only):\u003C\u002Fstrong>\u003C\u002Fp>\n\u003Cpre>\u003Ccode class=\"language-bash\">uv tool install git+https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskillspector.git\n# Update later: uv tool update skillspector\n\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Cp>If you plan to run \u003Ccode>skillspector mcp\u003C\u002Fcode>, install the MCP extra at install time:\u003C\u002Fp>\n\u003Cpre>\u003Ccode class=\"language-bash\">uv tool install 'skillspector[mcp] @ git+https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskillspector.git'\n\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Cp>\u003Cstrong>From source:\u003C\u002Fstrong>\u003C\u002Fp>\n\u003Cpre>\u003Ccode class=\"language-bash\"># Clone the repository\ngit clone https:\u002F\u002Fgithub.com\u002FNVIDIA\u002Fskillspector.git\ncd skillspector\n\n# Create and activate virtual environment\nuv venv .venv && source .venv\u002Fbin\u002Factivate\n# or: python3 -m venv .venv && source .venv\u002Fbin\u002Factivate\n\n# Install for production use\nmake install\n\n# Or install with development dependencies\nmake install-dev\n\u003C\u002Fcode>\u003C\u002Fpre>\n\u003Ch3>Docker (no Python required)\u003C\u002Fh3>\n\u003Cp>Run SkillSpector without installing Python by building it locally from the included [Dockerfile](Do\u003C\u002Fp>\n",1791762986054]