The Agent Skills Directory

[SAFE]: The skill operates locally using an Ollama model on a designated server (Mac Mini), ensuring file contents are not sent to external third-party services.
[SAFE]: Implements a mandatory 'dry-run, review, approve' workflow. The agent must summarize the plan and wait for explicit user confirmation before executing any moves.
[SAFE]: Includes a refusal list for sensitive directories such as .git, vault/, and Overleaf/, protecting system integrity and private data.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted file contents, which could theoretically contain instructions to influence the reorganization logic. This risk is mitigated by the mandatory manual review of the reorganization plan prior to execution.
[SAFE]: The CLI is a local script provided within the skill repository, minimizing external dependencies.

reorg