The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill facilitates the loading of laboratory deck layouts and resource states from external JSON files, creating an attack surface where maliciously crafted data could theoretically influence the agent's behavior if it interprets the content.
Ingestion points: references/resources.md (via Deck.load_from_json_file and plate.load_state).
Boundary markers: The instructions do not define explicit boundaries or warnings for the agent to ignore instructions embedded within these data files.
Capability inventory: The skill has capabilities for hardware control (e.g., Hamilton STAR, Opentrons OT-2), file system writes (to_csv, save), and network communication (Opentrons API).
Sanitization: There is no explicit evidence of JSON schema validation or content sanitization for imported laboratory states.
[EXTERNAL_DOWNLOADS]: The skill references the installation of the pylabrobot Python package from a public registry to enable its core functionalities.
Source: SKILL.md identifies pylabrobot as a required dependency.

pylabrobot