The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The README.md file and the automated security scan identify an installation command: curl -fsSL https://raw.githubusercontent.com/Alcyone-Labs/simple-logger/main/skills/install.sh | bash. This pattern downloads and executes a script directly from a remote source. While the repository belongs to the skill's author, this method bypasses manual review of the script's contents and creates a potential vector for executing unverified code if the remote source is compromised.
[COMMAND_EXECUTION]: In commands/opencode/simple-logger-usage.md, the workflow includes a step to execute skills/install.sh locally. This command execution is gated by a simple check for a string flag in the user input, which allows the skill to run shell scripts on the host system during normal operation.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its handling of user arguments in commands/opencode/simple-logger-usage.md which triggers the aforementioned command execution.
Ingestion points: The $ARGUMENTS variable is used to parse user intent and flags.
Boundary markers: None detected; the skill does not use delimiters or instructions to ignore embedded flags in the user input.
Capability inventory: The skill can execute shell scripts (install.sh) and read multiple reference files.
Sanitization: There is no sanitization or validation of the arguments; it simply checks if the string contains --update-skill. A malicious user could craft a prompt that tricks the agent into triggering the installation script unintentionally.

simple-logger-usage