Ontology Local

Typed knowledge graph for structured agent memory and composable skills. Use when creating/querying entities (Person, Project, Task, Event, Document), linkin...

This skill appears to implement a straightforward local ontology stored under memory/ontology and mostly matches its description; however: - I could not review the entire scripts/ontology.py (the provided file content was truncated). Inspect the full script before installing — grep for network, subprocess, eval/exec, or unusual file paths (look for 'requests', 'urllib', 'socket', 'subprocess', 'exec', 'eval'). - The _meta.json ownerId/version differ from the registry metadata; confirm the publisher and version are what you expect before trusting this package. - The skill intentionally stores 'secret_ref' strings rather than raw secrets, which is good, but ensure you never put actual secrets into the graph.jsonl or schema files. Treat the ontology files as potentially readable by other local processes. - Run the skill in a sandbox or with non-sensitive test data first. If you plan to integrate with other skills that dereference secret_ref values, audit those skills carefully for credential access. If you want, I can (1) scan the complete scripts/ontology.py for network/file-access patterns if you provide the full file, or (2) produce a quick checklist of exact strings to search for in the repository to verify there are no hidden endpoints.

Type: OpenClaw Skill Name: ontology-local Version: 1.0.0 The ontology skill bundle provides a structured knowledge graph system for agent memory, implementing standard CRUD operations and relationship management via a local JSONL file. The implementation in 'scripts/ontology.py' includes security best practices such as path traversal protection (resolve_safe_path) and safe YAML loading, while the schema definitions in 'references/schema.md' explicitly forbid the storage of raw secrets (passwords/tokens) in the graph.