The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external codebases and configuration files to identify security vulnerabilities. This creates an attack surface for indirect prompt injection if the untrusted data being analyzed contains malicious instructions targeting the agent.
Ingestion points: The agent reads external source code and configuration files during its 'Surface Identification' and 'Edge Case Probing' phases.
Boundary markers: There are no explicit instructions or delimiters provided to the agent to treat the analyzed code strictly as data or to ignore instructions embedded within comments or strings in the target files.
Capability inventory: The skill is granted access to the Read, Grep, and Glob tools to inspect the filesystem.
Sanitization: No sanitization or filtering of the ingested content is performed before it is processed by the agent.
[SAFE]: The skill contains various examples of high-risk patterns, including destructive system commands (rm -rf /), code injection vectors (eval, unserialize), and hardcoded credential examples. These are correctly documented within markdown code blocks in the language-specific reference files (e.g., references/lang-php.md, references/lang-ruby.md) as examples of anti-patterns and 'footguns' for educational purposes. They do not constitute active malicious behavior by the skill itself.

sharp-edges