The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill's Linux installation script suggests a usage pattern that involves piping remote content directly from a URL to the bash interpreter. Evidence: install.sh contains a usage comment suggesting curl -fsSL https://raw.githubusercontent.com/kznr02/OpenTester/main/opentester-skills/install.sh | bash.
[COMMAND_EXECUTION]: The skill provides the ability to execute arbitrary shell commands via its YAML-based DSL for test automation. Evidence: SKILL.md and REFERENCE.md document the exec action for CLI targets, which executes a user-generated command string.
[EXTERNAL_DOWNLOADS]: Installation scripts for both Windows and Linux download core skill components and examples from the author's GitHub repository at runtime. Evidence: install.sh and install.ps1 fetch files from github.com/kznr02/OpenTester.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it ingests untrusted data (local code changes and instructions) to generate executable tests.
Ingestion points: The agent reads local source code and Git diffs to understand changes as described in SKILL.md workflows.
Boundary markers: The YAML DSL provides structure, but there are no explicit prompt-level instructions to ignore malicious instructions embedded in the code being analyzed.
Capability inventory: Command execution is available through the run_case and run_project tools.
Sanitization: No sanitization or content validation is mentioned before the input is used to generate executable YAML steps.

opentester