The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs the user to install a CLI tool directly from a remote GitHub repository using package managers.
Evidence: Instruction to run uv tool install git+https://github.com/chhunlyyy/sentinel or pipx install git+https://github.com/chhunlyyy/sentinel in SKILL.md.
[INDIRECT_PROMPT_INJECTION]: The skill analyzes the content of local project files to determine configuration parameters, which could be exploited if those files contain malicious instructions aimed at the agent.
Ingestion points: Reads pubspec.yaml, package.json, and build.gradle from the project workspace to detect stack profiles.
Boundary markers: No explicit delimiters or warnings are used to separate the content of project files from the agent's instructions.
Capability inventory: The agent executes sentinel init and sentinel shared set commands using arguments derived from file content.
Sanitization: There is no evidence of sanitization or strict validation of the detected profile strings before they are incorporated into command-line arguments.

sentinel-setup