The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The source code for virtually the entire skill (approximately 60 files) is obfuscated using Pyarmor 9.2.4. This technique encrypts the logic and uses a proprietary runtime to execute it. This makes the skill's behavior completely unverifiable, as the actual operations performed on the user's system and data are hidden from audit and inspection.
[EXTERNAL_DOWNLOADS]: The skill's documentation and bootstrap script indicate that it automatically manages dependencies using the 'uv' tool. Since the logic for this process is obfuscated, it is impossible to verify which packages are being installed or the sources from which they are fetched, creating a risk of installing malicious third-party software.
[COMMAND_EXECUTION]: The skill utilizes various external CLI tools and system commands to perform document parsing and dependency management. The specific commands, shell invocations, and arguments are concealed within the obfuscated script files, preventing a proper audit of how the skill interacts with the underlying operating system.
[PROMPT_INJECTION]: The skill is designed to ingest and parse untrusted data from multiple formats (PDF, DOCX, HTML) and remote URLs, creating a significant attack surface for Indirect Prompt Injection.
Ingestion points: Input arguments processed by scripts/lyxy_document_reader.py and various reader modules.
Boundary markers: Implementation of delimiters cannot be confirmed due to the obfuscated nature of the parsing logic.
Capability inventory: The skill has extensive capabilities including file system access, network access, and command execution.
Sanitization: Verification of data sanitization or filtering logic is impossible due to obfuscation.

lyxy-document-reader