The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data read from terminal panes which creates a surface for indirect prompt injection.
Ingestion points: The agent reads content from other terminal panes using herdr pane read and herdr agent read as described in SKILL.md.
Boundary markers: There are no explicit instructions to use delimiters or ignore instructions within the read content.
Capability inventory: The skill has extensive capabilities including executing commands in other panes (herdr pane run), starting new agents (herdr agent start), and sending prompts to existing agents (herdr agent prompt).
Sanitization: No sanitization or validation of the content read from terminal panes is mentioned before it is processed by the agent.
[COMMAND_EXECUTION]: The skill is designed to execute various shell commands through the herdr CLI to control the terminal environment.
Evidence: SKILL.md provides numerous examples of using the herdr binary to mutate terminal layout, run processes, and manage remote sessions.

herdr