The Agent Skills Directory

[SAFE]: The skill operates with a human-in-the-loop security model, explicitly requiring user approval for all repairs and ensuring the agent remains under control.
[COMMAND_EXECUTION]: Local bash scripts are used to perform linter-like checks on workspace artifacts. These scripts utilize standard command-line tools (grep, awk, git, jq) for analysis.
[SAFE]: Repository modifications are scoped to the .woostack metadata directory and follow standard authoring conventions.
[COMMAND_EXECUTION]: Administrative git commands, such as 'git worktree prune', are executed locally to manage repository health as part of the doctor's diagnostic duties.
[SAFE]: The skill processes markdown files as data and uses robust parsing logic (jq, field extraction) to prevent indirect injection from influencing tool execution.
[SAFE]: No external network connections or unauthorized data exfiltration patterns were detected; all operations trace back to local vendor resources.

woostack-doctor