The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external source code using the Read and Grep tools, which creates an indirect prompt injection surface. Adversarial content within the audited codebase could attempt to influence the agent's behavior.
Ingestion points: The agent uses Glob, Grep, and Read to process untrusted source code, configuration files, and dependency manifests as outlined in Step 1 and Step 2 of the process.
Boundary markers: The skill contains an explicit 'Prompt Injection Safety Notice' section. This section instructs the agent to treat all code content as data (not instructions), ignore embedded directives in reviewed files, and never execute or evaluate the code it reads.
Capability inventory: The skill's allowed tools are limited to Read, Grep, and Glob. It does not have network access or file-write permissions, significantly limiting the potential impact of an injection.
Sanitization: The skill relies on these clear instructional boundaries to prevent cross-contamination between the data being analyzed and the agent's internal instructions.
[SAFE]: The skill includes numerous vulnerable code snippets and security-sensitive regex patterns (e.g., hardcoded password patterns, weak hashing functions). These are used solely for identification and remediation guidance within the context of a security audit and do not represent malicious behavior by the skill author.

owasp-top-10-web