{"full_name":"NVIDIA/SkillSpector","owner":"NVIDIA","name":"SkillSpector","shard":"1","url":"https://github.com/NVIDIA/SkillSpector","description":"Security scanner for AI agent skills. Detect vulnerabilities, malicious patterns, security risks, prompt injection, data exfiltration, and supply-chain risks in Claude Code, Codex, and MCP skills before you install them.","homepage":"https://docs.nvidia.com/skills/scanning-agent-skills","topics":["agent-security","agent-skills","agentic-ai","ai-security","claude-code","mcp","prompt-injection","security-scanner","security-tools","security-workflow","supply-chain-security"],"tag_pages":["agent-security","agent-skills","agentic-ai","ai-security","claude-code","mcp","security-scanner","security-tools"],"language":"Python","license":"Apache-2.0","avatar":"https://avatars.githubusercontent.com/u/1728152?v=4","created_at":"2026-03-21T00:28:43Z","pushed_at":"2026-11-14T04:49:48Z","metrics":{"stars":26197,"forks":2323,"watchers":95,"issues":404,"prs":709,"releases":22,"issues_open":79,"issues_closed":325,"prs_open":73,"prs_closed":71,"prs_merged":565,"discussions":4},"growth":{"day":{"stars":173,"forks":10,"base_day":"2026-11-15"},"week":{"stars":2011,"forks":240,"base_day":"2026-11-04"},"month":{"stars":6337,"forks":585,"base_day":"2026-10-17"}},"history":{"dates":["2026-09-06","2026-09-07","2026-09-08","2026-09-09","2026-09-10","2026-09-11","2026-09-12","2026-09-13","2026-09-14","2026-09-15","2026-09-16","2026-09-17","2026-09-18","2026-09-19","2026-09-20","2026-09-21","2026-09-22","2026-09-23","2026-09-24","2026-09-25","2026-09-26","2026-09-27","2026-09-28","2026-09-29","2026-09-30","2026-10-01","2026-10-02","2026-10-03","2026-10-04","2026-10-05","2026-10-06","2026-10-07","2026-10-08","2026-10-09","2026-10-10","2026-10-11","2026-10-12","2026-10-13","2026-10-14","2026-10-15","2026-10-16","2026-10-17","2026-10-18","2026-10-19","2026-10-20","2026-10-21","2026-10-22","2026-10-23","2026-10-24","2026-10-25","2026-10-26","2026-10-27","2026-10-28","2026-10-29","2026-10-30","2026-10-31","2026-11-01","2026-11-02","2026-11-03","2026-11-04","2026-11-10","2026-11-11","2026-11-12","2026-11-13","2026-11-14","2026-11-15","2026-11-16"],"stars":[14873,14942,14962,15005,15042,15058,15098,15193,15534,15623,15850,16114,16265,16534,16658,16777,16879,16960,17029,17172,17320,17427,17612,17747,17831,17973,18090,18166,18212,18296,18359,18506,18615,18930,19075,19240,19445,19553,19634,19706,19771,19860,20047,20125,20281,20366,20424,20567,20699,20803,20961,21074,21241,22001,22702,23028,23331,23513,23975,24186,25144,25340,25595,25759,25945,26024,26197],"forks":[1240,1250,1252,1257,1262,1263,1267,1273,1307,1318,1340,1368,1381,1408,1416,1427,1434,1444,1452,1468,1475,1490,1504,1522,1536,1551,1569,1573,1573,1580,1596,1613,1623,1651,1662,1681,1701,1708,1716,1722,1733,1738,1756,1770,1780,1788,1796,1805,1813,1826,1841,1850,1876,1925,1969,1991,2017,2034,2069,2083,2202,2221,2259,2280,2303,2313,2323],"watchers":[52,52,53,53,54,54,54,55,58,58,60,60,62,65,66,70,69,71,69,68,70,69,70,71,71,73,73,73,73,73,73,74,74,75,75,75,75,76,75,75,75,76,77,77,78,79,79,79,79,80,79,78,73,79,80,79,79,79,81,81,null,88,90,90,93,94,95],"issues_open":[null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,75,83,78,75,78,79],"prs_open":[null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,74,79,74,67,70,73]},"boards":{"daily":148,"weekly":170,"monthly":91,"monthly_rate":110},"same_owner":[{"full_name":"NVIDIA/DALI","name":"DALI","shard":"1","boards":{}},{"full_name":"NVIDIA/Isaac-GR00T","name":"Isaac-GR00T","shard":"1","boards":{}},{"full_name":"NVIDIA/Megatron-LM","name":"Megatron-LM","shard":"1","boards":{}},{"full_name":"NVIDIA/Model-Optimizer","name":"Model-Optimizer","shard":"1","boards":{}},{"full_name":"NVIDIA/NemoClaw","name":"NemoClaw","shard":"1","boards":{}},{"full_name":"NVIDIA/OpenShell","name":"OpenShell","shard":"1","boards":{"monthly":107,"monthly_rate":96}},{"full_name":"NVIDIA/SkillEvaluator","name":"SkillEvaluator","shard":"1","boards":{"daily":18,"weekly":138,"daily_rate":4,"weekly_rate":3,"monthly_rate":21,"daily_forks":18,"daily_watchers":189}},{"full_name":"NVIDIA/TensorRT-LLM","name":"TensorRT-LLM","shard":"1","boards":{}},{"full_name":"NVIDIA/apex","name":"apex","shard":"1","boards":{}},{"full_name":"NVIDIA/cosmos","name":"cosmos","shard":"1","boards":{}},{"full_name":"NVIDIA/cuda-samples","name":"cuda-samples","shard":"1","boards":{}},{"full_name":"NVIDIA/cudf","name":"cudf","shard":"1","boards":{}},{"full_name":"NVIDIA/cudnn-frontend","name":"cudnn-frontend","shard":"1","boards":{}},{"full_name":"NVIDIA/cutlass","name":"cutlass","shard":"1","boards":{}},{"full_name":"NVIDIA/garak","name":"garak","shard":"1","boards":{}},{"full_name":"NVIDIA/nvidia-docker","name":"nvidia-docker","shard":"1","boards":{}},{"full_name":"NVIDIA/open-gpu-kernel-modules","name":"open-gpu-kernel-modules","shard":"1","boards":{}}],"issues":[{"number":596,"title":"AE1 classifies opaque-by-type assets as `analysis-evasion` HIGH, discarding the ledger `reason_code` that distinguishes them from real coverage failures","comments":8,"created_at":"2026-09-20T18:59:25Z"},{"number":694,"title":"static_parse_limit on valid Rust, Python and POSIX shell source leaves files partially inspected","comments":7,"created_at":"2026-09-30T11:47:36Z"},{"number":37,"title":"Creates false positives on pure text-based content","comments":7,"created_at":"2026-06-13T09:09:09Z"},{"number":644,"title":"Static-pattern false positives found scanning public skills (P2 BOM, RA2 plist, EA3 OFL, plus design-level items)","comments":6,"created_at":"2026-09-26T18:32:13Z"},{"number":456,"title":"Add opt-in compact prompts mode to reduce LLM token usage","comments":6,"created_at":"2026-08-28T11:42:58Z"},{"number":271,"title":"Publish SkillSpector to PyPI for easier installation and integration","comments":6,"created_at":"2026-07-13T15:22:58Z"},{"number":121,"title":"Feature: MCP server registry scanning — complementary to skill scanning","comments":6,"created_at":"2026-06-20T21:20:02Z"},{"number":363,"title":"Oversized companion file causes SkillSpector to skip analysis of unrelated skill files (v2.8.2) (another malicious skill example that bypasses this scanner)","comments":5,"created_at":"2026-08-12T00:48:56Z"}],"discussions":[{"number":920,"title":"skillspector-rs: an optimized Rust port of the static analysis (no LLM yet)","comments":0,"category":"Show and tell","created_at":"2026-10-27T08:28:24Z"},{"number":871,"title":"Clarification requested: v2.13.0 release version verification","comments":0,"category":"General","created_at":"2026-10-18T15:36:19Z"},{"number":651,"title":"Missing category and findings when usellm is set","comments":0,"category":"General","created_at":"2026-09-28T06:48:48Z"},{"number":549,"title":"SkillSpector-TKV: A lightweight (~144KB) native port in TokenVector with GUI, offline-first scan, and parity rules","comments":0,"category":"Ideas","created_at":"2026-09-15T06:06:42Z"}],"latest_release":{"name":"SkillSpector v2.13.4","tag":"v2.13.4","published_at":"2026-11-04T22:40:20Z","body":"# SkillSpector 2.13.4\n\nThis maintenance release includes the v2.13.3 fixes and selected improvements to\nprovider compatibility, source inspection, and recovery of malformed AI responses.\n\n## Fixes\n\n- [#806](https://github.com/NVIDIA/SkillSpector/pull/806): provider compatibility\n  and structured responses for GPT-6.1 Sol and Claude Opus 5.5 across supported routes.\n- [#947](https://github.com/NVIDIA/SkillSpector/pull/947),\n  [#950](https://github.com/NVIDIA/SkillSpector/pull/950), and\n  [#973](https://github.com/NVIDIA/SkillSpector/pull/973): bounded parsing of shell,\n  PowerShell, JavaScript templates, Python strings, and Markdown documentation.\n- [#974](https://github.com/NVIDIA/SkillSpector/pull/974): inspect supported generated\n  shell helpers and archived sources without executing uploaded code. Dependencies\n  outside the uploaded skill and unprovable runtime behavior remain incomplete.\n- [#997](https://github.com/NVIDIA/SkillSpector/pull/997) and\n  [#999](https://github.com/NVIDIA/SkillSpector/pull/999): distinguish unsupported\n  syntax, unresolved parser boundaries, and exhausted resource limits.\n- [#1000](https://github.com/NVIDIA/SkillSpector/pull/1000): bounded TAR/GZIP\n  inspection with explicit accounting for omitted archive members.\n- [#979](https://github.com/NVIDIA/SkillSpector/pull/979),\n  [#998](https://github.com/NVIDIA/SkillSpector/pull/998), and\n  [#1007](https://github.com/NVIDIA/SkillSpector/pull/1007): preserve provider\n  failure details, bound retries, and distinguish outages from security findings.\n- [#1011](https://github.com/NVIDIA/SkillSpector/pull/1011): modern JavaScript parsing\n  and narrower classification of passive presentation assets and formatting metadata.\n- [#1014](https://github.com/NVIDIA/SkillSpector/pull/1014): shared recovery of\n  malformed AI assessments, including plain-text and individual-field requests.\n  Recovered findings remain visible. Unverifiable formatting-only responses carry\n  an explicit warning and retain unverified coverage; refusals, provider failures,\n  and unrelated inspection gaps retain their separate diagnostics.\n- The bounded advisory-count correction from\n  [#1010](https://github.com/NVIDIA/SkillSpector/pull/1010), plus required literal,\n  documentation, reference, and text-pattern prerequisites. The broader Python\n  docstring exemption from that PR is excluded.\n\nIntegration corrections preserve malformed single-object findings, avoid repeated\nshell-comment scans, and limit decoded signature metadata ownership to parser\nboundaries while retaining checks on unknown fields and decoded payloads. The\nmaintenance branch's scoped allowed-tools and output-file protections are retained.\nMarkdown list fences end with their containing list, so following prose does not\nextend a shell string. Source distributions contain the runtime package and public\npackage metadata.\n\n## Scope and upgrading\n\nThis release does not resolve every parser limitation or guarantee that every\nlarge skill completes. File, archive, retained-content, report, and time limits\nremain enforced. Additional container support can consume the shared byte budget\nand displace previously inspected members in oversized inputs; those omissions\nremain explicitly incomplete, and disappearing findings behind a limit are not\nsecurity fixes.\n\nDownstream consumers need compatible handling of `complete_with_caveats` and the\nformat-recovery warning policy. Updating the scanner alone does not change an\nevaluator's decision to block incomplete inspection. Some consumers also need\nseparate handling for passive BMP assets inside chart archives.\n\nRegenerate and review version-bound baselines for 2.13.4. Release artifacts are\npublished on GitHub as a wheel and source distribution; this workflow does not\npublish to PyPI.\n","is_prerelease":false},"readme":true,"readme_zh":false,"news":[{"title":"SkillSpector","url":"https://github.com/NVIDIA/SkillSpector","hn":"https://news.ycombinator.com/item?id=48509844","source":"Hacker News","lang":"en","date":"2026-06-12","points":49,"comments":4},{"title":"Скилл с GitHub может оказаться с сюрпризом","url":"https://vc.ru/id1890964/3220763-skaner-skillov-skillspector-ot-nvidia","source":"vc.ru","lang":"en","date":"2026-10-28"},{"title":"NVIDIA releases agent skill inspection tool. While AI budgets are increasing, only 29% of large companies can track spending across the entire organization | amu_lab AI/DX News ...","url":"https://note.com/bhrtaym/n/n759bee905ce3?hl=en","source":"note","lang":"en","date":"2026-10-26"},{"title":"Skillの安全スキャナは、鍵を盗む自作Skillを「SAFE」と判定した","url":"https://note.com/generativeai_new/n/na9bcb12f0f13","source":"note","lang":"zh","date":"2026-10-26"},{"title":"How to use SkillSpector: NVIDIA's free OSS for inspecting AI skills before installation, and how to read the scores","url":"https://note.com/opensourcelab/n/nb2b5ec56321e?hl=en","source":"note","lang":"en","date":"2026-10-14"},{"title":"Hottest cybersecurity open-source tools of the month: August 2026","url":"https://www.helpnetsecurity.com/2026/08/26/hottest-cybersecurity-open-source-tools-august-2026/","source":"Help Net Security","lang":"en","date":"2026-08-25"},{"title":"Building an Advanced AI Skill Security Auditing Pipeline with NVIDIA SkillSpector, LangGraph, YARA Rules, SARIF, and CI Policy Gates","url":"https://www.marktechpost.com/2026/08/04/building-an-advanced-ai-skill-security-auditing-pipeline-with-nvidia-skillspector-langgraph-yara-rules-sarif-and-ci-policy-gates/","source":"marktechpost","lang":"en","date":"2026-08-04"},{"title":"SkillSpector: NVIDIA’s open-source security scanner for AI agent skills","url":"https://www.helpnetsecurity.com/2026/08/03/skillspector-open-source-agent-skill-security-scanner/","source":"Help Net Security","lang":"en","date":"2026-08-02"},{"title":"What is NVIDIA SkillSpector? How to Safely Verify Local AI Agent Skills [Introduction Guide]","url":"https://note.com/zephel01/n/n61f9232d4437?hl=en","source":"note","lang":"en","date":"2026-06-21"},{"title":"NVIDIA SkillSpector Guide: Scanning AI Skills for Security Risks with Static Analysis and SARIF Reports","url":"https://www.marktechpost.com/2026/06/17/nvidia-skillspector-guide-scanning-ai-skills-for-security-risks-with-static-analysis-and-sarif-reports/","source":"marktechpost","lang":"en","date":"2026-06-16"},{"title":"How NVIDIA's SkillSpector works and How to add It to your skill PRs","url":"https://x.com/dani_avila7/article/2063336153630011728","source":"x","lang":"en","date":"2026-06-06"},{"title":"5 GitHub Finds That Exploded This Week (Free Tutor, NotebookLM Clone & More) #Shorts","url":"https://news.google.com/rss/articles/CBMiQkFVX3lxTE9sV3VuZk1WcGtRSTJyeEU0bmtDSW5zUE43SFRna04ySEkyQkx3RFFRenNjSHcyMlhUejVKRVdiaWpqQQ?oc=5","source":"YouTube","lang":"en","date":"2026-10-30"},{"title":"Detecting Vulnerabilities in Agent Skills with SkillSpector: From Green Checkmark to Real Security Judgment","url":"https://news.google.com/rss/articles/CBMisgFBVV95cUxPUG9tSVBHUnE5MXRNbl9nRW94cV9DV1FOZzN0c01uQi1CU2g5TkhNek96V1Nuamd4bmdvSTZLTGF4NDdUbHlJQTFYUy1RYXQ1WExHM0hSV3ZlVzBsNi12Smw1UXJIc21LQ0NSU0ppYjZqQUZteHZ1R1RuVjFzcGhIUnhLOV8wcGVlVERONG1qd2ZZbjBycGRNZ2thdzZwQmthSGNweTgzU1FCbUZkN0VWbU1R?oc=5","source":"Towards Data Science","lang":"en","date":"2026-07-22"},{"title":"英伟达开源SkillSpector：一款让Agent Skill安全无处遁形的扫描利器","url":"https://news.google.com/rss/articles/CBMiuwFBVV95cUxPc0tXakRaNGdjeTcycGVjdFVfcFdIT2Q3RzVBbVpMQmZfWDhScmdnUmY0S2RrazdyWW4yajlEeVhyaW5HWVZ5eVFiRXluSkZzeWFCekV0c01Td2o3TDE0NGdYTnZIT0RzMzkyN21nU09fUEVKbDVtbmEwSEg1TFY4UXE0eW1lTlBPZmJ5eGxhMWxQX1VCV3l0NDRGb2FHVDJPNmVBYUZTcGtoemx4ZW9oWUFpU3lvV2xhb2E0?oc=5","source":"积墨 AI","lang":"zh","date":"2026-06-26"},{"title":"谁是 Agent 最强守门员？首个 Agent 技能安全评测基准 SkillTrustBench 正式发布","url":"https://news.google.com/rss/articles/CBMieEFVX3lxTE9ZenNkOWZ6cmkwSS1ZbnZSX3pic1laWmcwQVVzMmNldDBPN3RLMWhVVlVDQ1UtcXBJQmlVTHhEM2g4THBKOGF1YXlPcFdCRWd4Nlh5X3BoTTcwS0NBZ1E3RWNZcWdwbGpnNl93elRyUFJXcDdJQUlxdQ?oc=5","source":"新浪财经","lang":"zh","date":"2026-06-16"},{"title":"26.1%的AI编程技能存在漏洞：NVIDIA开源SkillSpector能扫出什么？","url":"https://news.google.com/rss/articles/CBMivgFBVV95cUxPYWIxeFM3dlprYU9lMlUzRF8wOERFRFRZbmhEb196Q2hrcUI2emRKNHBHdDE1cmR4M1kwcDZZVkROdnpLUWhpdmFDeEhUSWszX1VUdERFak1SRXhnQjVtM280RGkwTjlZaDMtbjVxQVJ1TXpTMEtveXlyVEotMkNYSHotTHktV3NZZTB5VkFTV0YtZWJDSUJDMlBCQWk5akhGOTZENHRJSkxHVDJlQXVabHJCVjFGZFZmVWF2ZVdn?oc=5","source":"积墨 AI","lang":"zh","date":"2026-06-16"},{"title":"NVIDIA-Verified Agent Skills Provide Capability Governance for AI Agents | NVIDIA Technical Blog","url":"https://news.google.com/rss/articles/CBMiqwFBVV95cUxPd2drOXFmNVBGaHFIY2FOMFpoOUpiTXJ0YmFmcVJTa0VZSjR4bjBfNnc5QW1RUGJUMjYtalVfZGU0YV9ZenRucXBGM1dJZzNxY2hJcXdKdVpuWGxsZUh6UGtZV0JDcTJqRGYwcHA0RlB3N2dTR2JLSGlETUhnUkhxRnVaN2Nfd2plRmJCVTNzRnZFMmZYSmh1R05raHh5d1NVa01XUllWZnVDMEE?oc=5","source":"NVIDIA Developer","lang":"en","date":"2026-05-19"},{"title":"科技大厂疯抢“龙虾”，AI新战场开启？_IT界那些事儿-龙虾开发者社区","url":"https://news.google.com/rss/articles/CBMibkFVX3lxTFB3LXBkTkY1clMwcWZxUUhWNE9VU2FQQzlUZ0hPdjFQeXNrVzhUY09meWtzUVEwQWtkZENDSl9DTVpuYVd5UmJMTVdiTFhuWWpoOEZ4S2FVc2VWbFBBNDV3Wkcxc29Bb09Ud2FWcG93?oc=5","source":"CSDN","lang":"zh","date":"2026-03-10"}],"stale":false,"snapshot_at":"2026-11-16T09:23:14Z","history_start":"2026-09-06"}