The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process original and rewritten text files provided as before.md and after.md.
Ingestion points: Input files before.md and after.md.
Boundary markers: The skill performs structural comparison to identify deviations from the original text, although it does not provide explicit instruction-boundary markers within the analyzed content.
Capability inventory: The skill uses scripts/validate.js to read files using node. The script utilizes fs.readFileSync for comparisons and does not have capabilities for network operations, file writing, or arbitrary command execution based on content.
Sanitization: The validator uses regular expressions to identify and count specific writing patterns and does not execute or interpret the content of the text files.
[SAFE]: The skill's logic is self-contained within the provided repository, including the AIDetector and AIDetectorValidate modules.
[SAFE]: The primary purpose of the skill is defensive auditing and content verification, and it does not request high-privilege access or perform external communication.

preservation-verifier