The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill does not involve downloading or executing code from external or untrusted remote servers. All Python automation logic is provided in local scripts included in the package.
[DATA_EXFILTRATION]: No network-based exfiltration patterns or unauthorized data transmission were found. The skill actively advises the agent to use the --factory-startup flag to avoid running user-defined add-ons that might perform unexpected network calls.
[COMMAND_EXECUTION]: All command execution is confined to the Blender Python API (bpy) for legitimate 3D production tasks such as geometry manipulation, audit reporting, and rendering. File operations are limited to saving renders and JSON reports to specified local paths.
[INDIRECT_PROMPT_INJECTION]: While the skill processes external 3D mesh data, it does so using objective geometric audits and standard Blender data access. There are no patterns suggesting that data attributes are interpreted as instructions for the agent.
[DYNAMIC_CONTEXT_INJECTION]: There is no use of the dynamic context injection syntax (!command) within the SKILL.md file.
[OBFUSCATION]: The skill's instructions, reference documents, and scripts are provided in clear, human-readable text without any signs of deceptive encoding or hidden characters.

blender-expert