The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute the claude CLI command with dynamically generated arguments. Specifically, it builds a command string claude --bg --name "<descriptive name>" "<handoff summary>" where the summary is derived from the current conversation context.
[PROMPT_INJECTION]: The skill transforms untrusted conversation data into a shell command, creating a surface for indirect prompt injection. * Ingestion points: Current conversation history and user arguments (SKILL.md). * Boundary markers: None used to encapsulate the summary within the command string. * Capability inventory: Shell command execution via the claude CLI (SKILL.md). * Sanitization: Includes a manual prompt instruction to redact sensitive information like API keys or passwords.
[DATA_EXPOSURE]: Conversation context is passed as a command-line argument. While the skill advises redacting sensitive info, this relies on the model's reliability in identifying all secrets before command execution.

claude-handoff