Scan external SKILL.md files for prompt injection and red-flag phrases.
npx skills add https://github.com/cris-m/flopsy --skill skill-security
Please help me install this Agent Skill. Skill: skill-security Source: https://github.com/cris-m/flopsy/tree/main/src/team/templates/skills/skill-security Command: npx skills add https://github.com/cris-m/flopsy --skill skill-security
Detect prompt injection, malicious SKILL.md files, poisoned config files, and social engineering attacks targeting the agent. Use when loading external skills, processing untrusted content, or reviewing configuration files.
Run a safety scan on the new SKILL.md and any accompanying scripts before enabling the skill.
High-intent search queries and answers about installing and using this skill.
To detect prompt injection in external skills, run a safety scan on the SKILL.md file and accompanying scripts. This validates frontmatter, identifies anomalous content like Unicode steganography, and flags risky patterns before enabling the skill.
Unicode steganography in untrusted content hides malicious payloads using invisible or anomalous characters. Security scanning detects these hidden anomalies during external skill evaluation, preventing covert data exfiltration and prompt injection attacks.
Review configuration artifacts for malicious content by applying safe-handling protocols during evaluation. This involves scanning for poisoned config files, validating frontmatter, and detecting red-flag phrases that indicate social-engineering attempts.
Yes, you can validate frontmatter and scan for red-flag phrases. The skill implements anomaly detection and a scanning protocol to identify risky patterns in skill definitions, providing actionable guidance for safe evaluation.
The best way to guard an agent from social-engineering attempts is to scan external SKILL.md files for red-flag phrases and anomalous content. This prevents data exfiltration by applying safe-handling protocols during skill loading.
jmeier1963Scaffold an ΩmegaWiki skeleton from raw sources with YAML frontmatter templates.
CraftOS-devAnalyze AI agent skill source code for malicious patterns and risk levels.
SergeiGolosDetect and prioritize security vulnerabilities across dependencies, secrets, code patterns, and configurations.
charleshall888Detect missing trigger phrases in SKILL.md frontmatter for regression tests.
