The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install a CLI tool using npx -y @mvanhorn/printing-press-library and go install github.com/mvanhorn/printing-press-library/..., which fetches code from external, untrusted repositories.
[REMOTE_CODE_EXECUTION]: By requiring the download and execution of the uspto-tsdr-pp-cli binary from an unverified GitHub repository and NPM package, the skill facilitates the execution of remote code on the host environment.
[DATA_EXFILTRATION]: The CLI includes a --deliver webhook:<url> argument that allows sending the results of trademark queries to an arbitrary external URL. Additionally, the feedback mechanism can be configured to automatically exfiltrate data via the USPTO_TSDR_FEEDBACK_ENDPOINT environment variable.
[COMMAND_EXECUTION]: The skill relies on the execution of the uspto-tsdr-pp-cli tool via the Bash tool, passing user-provided arguments that could lead to arbitrary command execution if the binary or input handling is compromised.
[CREDENTIALS_UNSAFE]: The documentation suggests storing sensitive USPTO API keys in an environment variable (TSDR_APIKEY_HEADER) or a local configuration file (~/.config/tsdr-pp-cli/config.toml), making them accessible to other local processes.
[PROMPT_INJECTION]: The skill processes untrusted data from the external USPTO TSDR API without boundary markers or sanitization while having access to the Bash tool, which presents an attack surface for indirect prompt injection.

pp-uspto-tsdr