The Agent Skills Directory

[PRIVILEGE_ESCALATION]: The skill's installation instructions for Debian/Ubuntu systems recommend the use of sudo dpkg -i, which grants the agent or user administrative privileges to install the package.
[INDIRECT_PROMPT_INJECTION]: The skill enables the agent to fetch and display the contents of emails and threads, which are untrusted external inputs. These inputs could contain malicious instructions intended to manipulate the agent's behavior.
Ingestion points: The commands paw-mail emails list, paw-mail emails fetch, paw-mail emails body, and paw-mail threads get all ingest data from remote mail servers into the agent's context.
Boundary markers: There are no instructions provided to the agent to treat email content as untrusted or to wrap it in specific delimiters to prevent instruction execution.
Capability inventory: The agent is granted permission to use Bash to execute paw-mail subcommands, including destructive actions like emails delete, emails move, and accounts delete.
Sanitization: The skill does not describe or implement any sanitization, filtering, or validation steps for the retrieved email data before processing.
[EXTERNAL_DOWNLOADS]: The installation section provides instructions to download and install the CLI tool from the author's GitHub repository and other external package managers like Homebrew and Arch User Repository.
Sources include github.com/pawpair/paw-mail-cli and pawpair/paw-mail-cli tap.
Installation methods include nix profile install, brew install, yay -S, and dpkg -i.

paw-mail-usage