The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external sources, creating a potential attack surface for instructions embedded in retrieved content.
Ingestion points: The skill retrieves reference images and visual data using browser tools and host image search as described in references/astra-workflow.md to guide modeling decisions.
Boundary markers: No explicit boundary markers or instructions to treat external data as untrusted are present in the guidelines.
Capability inventory: The skill has high capability, including generating Python scripts, writing files, and invoking platform-specific Blender validation and rendering tools.
Sanitization: There is no evidence of sanitization or filtering for data retrieved from external sources before it is processed or used to influence script generation.
[EXTERNAL_DOWNLOADS]: The skill references and downloads content from external web locations.
Fetches 3D materials and assets from Poly Haven, a well-known community resource for high-quality assets.
References official technical documentation from OpenAI's developer guides.
[DYNAMIC_EXECUTION]: The skill's primary operation involves the automated generation and execution of procedural logic.
Generates Python source code to drive Blender's internal modeling engine and Geometry Nodes systems as specified in SKILL.md.
Instructions encourage the creation of durable, reproducible scripts for scene assembly.

blender-procedural-workflow