The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The script scripts/p_shape.py generates Python protocol code by interpolating user-provided strings (such as labware types, pipette models, and deck slots) directly into f-string templates. This represents an indirect injection surface where a malicious user could provide inputs that alter the logic of the generated script.
Ingestion points: User-supplied arguments for source_slot, dest_slot, tiprack_slot, source_labware_type, dest_labware_type, tiprack_type, pipette, pipette_mount, and source_well are processed in scripts/p_shape.py.
Boundary markers: None. The script does not use escaping or delimiters to isolate user strings within the generated code.
Capability inventory: The skill generates Python code intended for the Opentrons Robot API; it does not execute the generated code itself.
Sanitization: The script validates that the volume_ul is positive and within the pipette's physical capacity, but it does not perform sanitization on the string-based inputs used in the code generation templates.

opentrons-workflow