The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script build_evidence_inventory.py to automate the detection of matching files and text. This is a legitimate part of the skill's auditing functionality.
[COMMAND_EXECUTION]: The build_evidence_inventory.py script uses subprocess.check_output to call git commands (rev-parse, rev-list, hash-object). This is used to check the project's Git history for historical evidence of copied files, which is consistent with the stated purpose of the skill. The inputs to these commands are derived from the local site root and file paths, not untrusted remote sources.
[SAFE]: The script performs file system read operations on the target site and reference directories to generate SHA-256 hashes and analyze text shingles. These operations are scoped to the directories provided by the user in the command arguments.

audit-reference-originality