The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides a CLI tool (scripts/ontology.py) for managing a graph-based memory system. It performs standard file operations (read/append) on local JSONL and YAML files.
[DATA_EXPOSURE]: The skill implements a robust Credential entity type that explicitly forbids properties like password, secret, token, or api_key. It enforces the use of indirection through a secret_ref to prevent accidental credential leakage in the graph data.
[PATH_TRAVERSAL]: The Python implementation includes a resolve_safe_path function that resolves user-provided paths and ensures they remain within the workspace root, effectively mitigating directory traversal attacks.
[REMOTE_CODE_EXECUTION]: While the skill uses external packages like PyYAML, it uses yaml.safe_load for parsing schema files, protecting against unsafe deserialization. There are no patterns involving remote code downloads or dynamic execution of untrusted input.

ontology