{"categories":[{"label":"Security","url":"https://skillfed.io/packages/category/security/2"}],"enrichment":{"capability":"LLM Guard provides input and output scanning for Large Language Models to detect and prevent prompt injection, data leakage, harmful language, and other security threats.","skillfed_tags":["llm-security","prompt-injection","pii-detection"],"use_cases":["Scan user prompts before sending to a production LLM API to block prompt injection and data exfiltration attempts","Anonymize personally identifiable information in user inputs and deanonymize model outputs to comply with privacy requirements","Detect and filter toxic, biased, or harmful language in both prompts and LLM responses","Validate LLM outputs for code injection, malicious URLs, and factual consistency before returning to users","Enforce organizational policies by banning specific topics, competitors, or substrings in LLM interactions"],"what_it_does":"LLM Guard is a security toolkit that scans both user inputs (prompts) and model outputs to detect and mitigate threats in Large Language Model interactions. It offers multiple scanning strategies including prompt injection detection, PII anonymization, secret detection, toxicity analysis, and output validation for harmful content, code injection, and factual consistency. The package integrates with popular LLM APIs and is designed for production deployment.\n\nThe package depends on 12 runtime libraries spanning NLP (nltk, transformers, tiktoken), data anonymization (presidio-analyzer, presidio-anonymizer, faker), and ML inference (torch). However, the project is now abandoned\u2014its repository is archived and the last commit was 2026-07-08, meaning no new features, bug fixes, or security patches will be released. For teams considering adoption, this means relying on a static codebase for an inherently security-sensitive tool.","worth_installing":"No\u2014not for new projects. While the package is feature-rich and has low install friction, its abandoned status (archived repo, no commits since 2026-07-08) is disqualifying for security-sensitive work. An unmaintained firewall cannot be trusted to defend against evolving threats. Consider it only if you have a specific, time-bounded use case and can fork and maintain it yourself."},"id":"llm-guard","links":{"html":"https://skillfed.io/packages/llm-guard","md":"https://skillfed.io/packages/llm-guard.md","pypi":"https://pypi.org/project/llm-guard/"},"maintenance":{"status":"abandoned"},"meta":{"latest_release":"2025-05-19","license_spdx":null,"license_treatment":"unclear","name":"llm-guard","python_support":"capped_below_current","summary":"LLM-Guard is a comprehensive tool designed to fortify the security of Large Language Models (LLMs). By offering sanitization, detection of harmful language, prevention of data leakage, and resistance against prompt injection attacks, LLM-Guard ensures that your interactions with LLMs remain safe and secure."},"popularity":{"monthly_downloads":211788,"position":9474,"tier":"top_15000"},"security":{"n_vulnerabilities":0},"version":"0.3.16"}