688. flagged artifacts, skill names and campaigns
ShieldMap · AI security, mapped
What attackers slip into the AI stack.
- Campaigns
- 14
- Malicious packages
- 161
- Look-alike names
- 143
- Flagged extensions
- 347
- Skill name patterns
- 27
Packages published as malware, packages named to pass for AI libraries, AI extensions a store removed or a security report flagged, and the names of malicious agent skills, with the campaigns behind them. Each bar is one artifact, or one name pattern for a skill; a campaign's bar is as long as the packages it hit.
- Campaigns14
- Malicious packages161
- Look-alike npm packages79
- Look-alike PyPI packages64
- VS Code Marketplace extensions226
- Open VSX extensions51
- Browser extensions70
- Malicious agent skills27 name patterns
- 2026-11-17Microsoft agent-framework PyPI compromise1 affected
- 2026-10-21LiteLLM release pipeline compromise (1.107.0)1 affected
- 2026-06-17@mastra npm scope compromise7 affected
- 2026-05-19Microsoft durabletask PyPI compromise1 affected
- 2026-05-12guardrails-ai 0.10.1 malicious release1 affected
- 2026-05-11Mistral AI SDK compromise4 affected
Check your machine
Finds malicious packages and agent skills in your agent and MCP configs, installed skills, global npm packages, the npx and uvx caches, uv and pipx tools, Claude Code and Cursor. It runs offline and sends nothing anywhere.
pip install "shieldpi>=0.10.0"
shieldpi agent-config . --installedMalicious packages come from ShieldMap's advisory snapshot, look-alike names from the OpenSSF Malicious Packages project, extensions from the VS Code Marketplace and Open VSX removal lists and MalExt Sentry, and skill names from the indicators the shieldpi CLI ships, taken from one curated IOC list and not a full list. Data as of 2026-11-22. Report a correction to [email protected].