The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill instructs users to install the Tempo CLI by piping a remote script directly into the shell (curl -sSL https://tempo.im/install.sh | sh). This is a dangerous practice that allows for arbitrary code execution from a third-party server.
[EXTERNAL_DOWNLOADS]: The skill performs an unverified download of an installation script from https://tempo.im during the setup process.
[COMMAND_EXECUTION]: Internal scripts (check_stats.js, submit_ad.js, topup_ad.js) use the execSync function to execute tempo CLI commands. While the skill attempts to validate some inputs using regular expressions, this pattern increases the risk of command injection if validation is bypassed.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its data ingestion process.
Ingestion points: The skill reads product information from local files such as README.md, package.json, and Cargo.toml in SKILL.md (Step 1).
Boundary markers: No delimiters or safety instructions are provided to the agent to distinguish between data and embedded instructions in these files.
Capability inventory: The skill can execute network requests and trigger financial transactions (MPP payments) through its submission and top-up scripts.
Sanitization: The validate_ad.js script checks for field length and basic URL format but does not perform any sanitization for malicious natural language instructions.

ad-creator