The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a shell script block for environment validation. This script strictly performs local file existence checks for 11 specific documentation files using the standard shell test operator '-f'. No external binaries or network-linked commands are executed.\n- [DATA_EXFILTRATION]: The skill reads project documentation from 'docs/reverse-engineering/' and writes artifacts to '_portable-extract/'. No patterns of data exfiltration, such as network transmissions or access to sensitive user credentials or SSH keys, were found.\n- [PROMPT_INJECTION]: The skill processes content from documentation files during extraction. Ingestion points: docs/reverse-engineering/*.md. Boundary markers: Absent. Capability inventory: Limited to local file read and write. Sanitization: Absent. While this creates a surface for indirect prompt injection, the risk is mitigated by the lack of high-privilege execution capabilities or network access.\n- [SAFE]: The skill instructions, metadata, and logic follow security best practices. There are no signs of obfuscation, hardcoded secrets, or persistence mechanisms.

portable-extract