The Agent Skills Directory

[COMMAND_EXECUTION]: The skill relies on executing various shell commands via the clay CLI tool to perform its operations, including environment verification (clay whoami) and running data enrichment actions (clay workflows actions test).
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data (personal email addresses) and interpolates them into shell commands, creating a vulnerability surface.
Ingestion points: Personal email addresses are collected from the user or external lists during Step 1.
Boundary markers: There are no explicit instructions for the agent to use specific delimiters or escape sequences when inserting these emails into the JSON payloads used in CLI commands.
Capability inventory: The skill utilizes subprocess calls to the clay CLI to interact with third-party providers, which could be exploited if an input email contains shell metacharacters.
Sanitization: The instructions do not define sanitization or validation logic for the input strings before they are executed as part of a shell command.
[DYNAMIC_EXECUTION]: The skill includes instructions for the agent to dynamically discover action keys and schemas at runtime using shell pipelines (e.g., piping clay workflows actions list into jq), which involves the assembly and execution of commands based on external tool output.

convert-personal-email-to-work-email