The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands to interact with the platform-specific CLI tools.
Evidence: The skill instructs the agent to run plannotator-tui herdr open and provides a fallback herdr plugin pane open command.
Context: These commands are used to manage the human review interface and are aligned with the skill's primary purpose of facilitating document annotation.
[INDIRECT_PROMPT_INJECTION]: The skill operates by ingesting external data from files and user feedback, creating a potential surface for indirect prompt injection.
Ingestion points: Markdown documents written to the local filesystem and human feedback messages received in the chat.
Boundary markers: The skill specifies a structured feedback format (e.g., "## 1. (line 12)") to help the agent distinguish feedback items.
Capability inventory: The agent has the capability to execute shell commands for the Herdr environment and access local files within the project directory.
Sanitization: No explicit sanitization is performed on the file content or the feedback, as the agent is expected to interpret them as instructions for refining its plan.

plannotator-tui