The Agent Skills Directory

[SAFE]: The skill implements standard Blender extension management functionality using vendor-provided libraries dcc_mcp_core and dcc_mcp_blender.- [SAFE]: Actions such as modifying user preferences and performing addon installations are consistent with the skill's primary purpose and metadata.- [INDIRECT_PROMPT_INJECTION]: The skill provides an attack surface for indirect prompt injection by processing external package archives.
Ingestion points: The package object containing archive_path in scripts/plan_extension_install.py and scripts/install_extension.py.
Boundary markers: Not explicitly defined in the wrapper scripts; functionality is delegated to the dcc_mcp_blender library.
Capability inventory: Filesystem modifications for addon installation and modification of user preferences as declared in SKILL.md.
Sanitization: The tools.yaml schema includes a sha256 field for checksum verification, which allows for integrity checking of the downloaded packages before installation.

blender-extensions