The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses several shell scripts (scripts/clock.sh, scripts/debrief-gate.sh, scripts/guard-kata.sh, scripts/streak) to manage session state, enforce kata rules, and calculate attendance streaks. These scripts operate on local files within the ~/.sensei directory and the user-defined dojo directory.
[DYNAMIC_CONTEXT_INJECTION]: The SKILL.md file uses the !command syntax to inject dynamic information such as the current date, the learner's configuration, session logs, and attendance streaks. These commands are benign and focused on providing context for the learning session.
[REMOTE_CODE_EXECUTION]: While the skill includes scripts to run tests (assets/templates/python/check and assets/templates/ts/check), these are used to execute the learner's own code locally during the kata phase. They do not fetch or execute code from remote or untrusted sources.
[DATA_EXFILTRATION]: There are no network operations detected in the skill scripts or instructions. All data, including the learner model, logs, and progress tracks, is stored locally in the ~/.sensei/ directory. The skill specifically targets the local development environment (dojo repo).

sensei