The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructions and scripts recommend installing a Python dependency directly from a personal GitHub repository (https://github.com/mcmero/labarchives-py). This source is not associated with an official organization or a well-known service registry.
[REMOTE_CODE_EXECUTION]: By instructing users to run pip install git+https://github.com/mcmero/labarchives-py, the skill facilitates the execution of unversioned code from an external third-party source during installation.
[COMMAND_EXECUTION]: The provided Python scripts (setup_config.py, notebook_operations.py, entry_operations.py) perform sensitive system-level tasks, including creating local configuration files, writing notebook data backups to the filesystem, and performing network operations via API requests.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection because it ingests and processes untrusted data from external Electronic Lab Notebooks (ELN).
Ingestion points: External data enters the agent context through the users/user_access_info and notebook_backup API methods invoked in scripts/notebook_operations.py.
Boundary markers: Absent. The skill does not use specific delimiters or instructions to protect the agent from executing commands hidden within scientific data or entry text.
Capability inventory: The skill possesses capabilities for filesystem writing (notebook backups), network transmission (API requests via the requests library), and shell command execution.
Sanitization: While the scripts use standard libraries for XML and JSON parsing, they do not perform validation or filtering to detect or strip malicious instructions from the retrieved notebook content.

labarchive-integration