The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted content from external sources which could potentially contain malicious instructions or manipulated data. \n- Ingestion points: Reads data from $HOME/todo.txt and GitHub project items, including issue titles and task descriptions (SKILL.md). It also follows routing tags (@route todo:) found within the todo.txt content to read additional local files. \n- Boundary markers: Absent. The instructions do not provide delimiters or clear guidance to help the agent distinguish between administrative instructions and the data being processed. \n- Capability inventory: The skill is capable of writing files to the local directory (regroup/) and executing command-line tools such as gh and tuxedo (SKILL.md). \n- Sanitization: The skill enforces a specific structural output format (phases and tasks), providing a basic level of control over the resulting report, but does not include explicit content filtering or sanitization of the source text. \n- [COMMAND_EXECUTION]: The skill utilizes external command-line utilities to retrieve project information. \n- Evidence: Executing gh project item-list and gh issue list to fetch GitHub data (SKILL.md). It also supports an optional tuxedo lsa command for parsing todo files if the utility is available in the environment.

do-regroup