The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data by scanning the user's local source code files to identify API routes and data structures. While this creates a theoretical injection surface where malicious code comments or identifiers could attempt to influence agent output, it is a necessary part of the skill's primary function.
Ingestion points: The skill uses find and grep commands to scan file content in the user's project directory (e.g., in SKILL.md).
Boundary markers: No specific boundary markers or "ignore embedded instructions" warnings are used when processing scanned code.
Capability inventory: The skill performs filesystem reads and writes to generate test files. It does not perform network operations or privilege escalation.
Sanitization: The instructions do not specify sanitization for data extracted from source files before generating test templates.
[SAFE]: The shell commands used for route detection are standard local utilities (find, grep, sed, sort) and are restricted to the local environment without network or administrative requirements.
[SAFE]: The example test files in references/example-test-files.md include common security testing payloads (such as SQL injection and XSS strings). These are intended solely for validating the security of the target API and are not directed at the agent or host environment.

api-test-suite-builder