The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes a local Python script scripts/assets.py to handle all file management tasks. This script performs operations such as scanning the filesystem, moving files for archival purposes, and updating metadata in JSON files. While these are intended functions, the execution of a script with filesystem write access is a notable capability.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it reads the full content of text-based files (e.g., .md, .txt, .script) to perform keyword searches. If a file in the managed directory contains malicious instructions, they could be ingested by the agent during a search operation.
Ingestion points: Files located in the outputs/ directory, specifically those identified as text, script, or note types.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to ignore or isolate instructions found within the content of these files.
Capability inventory: The agent can move files, create directories, and modify tags through the provided Python script.
Sanitization: The assets.py script reads and returns file content without sanitization or escaping, which is then processed by the agent.

asset-manager