The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documents the use of the -c/--command flag with the add and edit subcommands. This flag allows the agent to embed specific shell commands into task metadata which are then executed when the start command is invoked. This constitutes a deliberate command execution capability.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection because it ingests data from external sources and has significant execution capabilities.
Ingestion points: The import and sync subcommands fetch data (titles, descriptions, and comments) from GitHub issues and Pull Requests into the agent's task management system.
Boundary markers: The instructions do not define boundary markers or provide warnings to the agent to disregard instructions that may be embedded within the imported GitHub data.
Capability inventory: The skill can execute arbitrary shell commands via the start -c feature, modify the local file system through task metadata updates, and manage the task stack.
Sanitization: There are no documented procedures for sanitizing, escaping, or validating the external content retrieved from GitHub before it is processed or displayed to the agent.

nytid-todo