The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to configure a CLI tool (fcsa) that supports a pluggable OCR backend. This backend can be configured to execute external commands (e.g., fc_ocr from ellmos-filecommander). This is a documented feature and the skill provides clear instructions for safe setup.
[INDIRECT_PROMPT_INJECTION]: The skill defines a workflow for scanning and categorizing files based on templates and content. This creates an attack surface where maliciously named or crafted files could attempt to influence the agent's sorting logic. The skill mitigates this risk by relying on structured configuration files (categories-definitions.json) and mandatory user review of dry-run results.
[DATA_EXPOSURE]: The skill identifies and protects sensitive data by explicitly instructing the agent to reject scan paths containing .PRIVAT or CREDENTIALS segments. This prevents accidental exposure or modification of sensitive credentials or private files during automated sorting.
[SAFE]: The skill emphasizes a 'safety-first' model: a mandatory dry-run is required for every new configuration, and the deletion action is hard-coded to a fail-closed 'trash' mode unless multiple explicit configuration flags and user approvals are provided.

file-collect-sort-action