The Agent Skills Directory

[SAFE]: The skill is entirely composed of documentation, templates, and evaluation data for software testing methodologies. It does not contain executable scripts or commands that interact with the host system or external networks.
[OBFUSCATION]: Static analysis identified Unicode steganography patterns in references/tours.md. This is a false positive; the file explicitly documents a 'Bad-data tour' and provides the Unicode Bidi Override (RLO) character and Zero Width Joiner (ZWJ) sequences as literal examples for testers to use when verifying how a system handles pathological inputs. These characters are used correctly for educational and structural purposes within the domain of software testing.
[DATA_EXPOSURE]: No hardcoded credentials, sensitive file paths, or data exfiltration patterns were found. Examples provided in templates (such as promo codes or mock user data) are clearly illustrative and benign.
[REMOTE_CODE_EXECUTION]: There are no patterns involving external package installations, remote script downloads, or dynamic execution of untrusted code. The skill maintains a documentation-only scope.
[INDIRECT_PROMPT_INJECTION]: While the skill defines how an agent should assist in planning tests based on user-provided feature descriptions, it does not provide any automated processing or tool-chaining that would expose the system to execution of instructions embedded in that data. The interaction remains within the bounds of standard assistant behavior.

exploratory-testing