The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires downloading and installing an external, unverified Python package labarchives-py from an individual GitHub repository (https://github.com/mcmero/labarchives-py). This represents a supply chain risk, as the source repository is untrusted and outside official package registries or verified vendor organizations.
Evidence: Found in SKILL.md, scripts/entry_operations.py, scripts/notebook_operations.py, and scripts/setup_config.py recommending installation via git clone https://github.com/mcmero/labarchives-py or pip install git+https://github.com/mcmero/labarchives-py.
[INDIRECT_PROMPT_INJECTION]: The skill processes data retrieved from the LabArchives REST API, which represents an ingestion surface for potential indirect prompt injection if an attacker can manipulate notebook contents or user entries.
Ingestion points: scripts/notebook_operations.py and scripts/entry_operations.py call API methods to list notebooks and fetch entry/backup content.
Boundary markers: Absent. There are no delimiters or explicit instructions telling the agent to ignore any natural language instructions embedded within the notebook contents.
Capability inventory: The skill includes local file writing capabilities (saving backups to disk in scripts/notebook_operations.py) and outbound HTTP requests via requests.
Sanitization: Absent. Content retrieved from the API is parsed using standard XML/JSON libraries without filtering or sanitizing against embedded instructions.

labarchive-integration