The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Fetches the bento-pp-cli binary using npx from the @mvanhorn/printing-press-library package on NPM or via go install from the author's GitHub repository (github.com/mvanhorn/printing-press-library). These are official resources provided by the skill's author.
[COMMAND_EXECUTION]: The skill executes the bento-pp-cli tool to perform account management, statistical analysis, and data transformation tasks.
[DATA_EXFILTRATION]: Provides a --deliver webhook:<url> flag that allows routing command output to a remote URL. This is a functional feature for automation, though it could be used to send data externally if misconfigured.
[PROMPT_INJECTION]: The skill ingests untrusted data from external files (CSV/JSON) which is then returned to the agent's context, creating an indirect prompt injection surface.
Ingestion points: Processes files via --in, --emails-from, --from, and --shipped-csv flags as documented in SKILL.md.
Boundary markers: Absent; there are no specific instructions or delimiters to isolate data content from instructions when output is returned to the agent.
Capability inventory: The skill uses the Read Bash tool to execute CLI commands and handle results, which may include sensitive marketing data.
Sanitization: While the tool performs 'hygiene' scrubs on marketing data (e.g., email validation), it does not implement sanitization to prevent the agent from interpreting data as instructions.

pp-bento