'Understand architecture, find code, and trace callers or execution flows in an unfamiliar repository.'
Install
npx skillscat add ontograph/ontoindex/ontoindex-cursor-integration-skills-ontoindex-exploring Install via the SkillsCat registry.
Read ontoindex-navigation first (freshness gate and failure path).
Find and explain code
Start with a concept, then inspect the returned symbol identity:
search({action: "semantic", repo: "<repo>", query: "<concept>", limit: 5})
gn_explore({repo: "<repo>", query: "<concept>", depth: "balanced"})
inspect({action: "context", repo: "<repo>", target: "<symbol>", include_content: true})
gn_find_related({repo: "<repo>", symbol: "<symbol>", maxItemsPerCategory: 10})Use returned paths and identities to read current source and resolve ambiguous symbols.
Trace the entry point, calls, state changes, and output; cite the evidence for each connection.
Separate graph facts from hypotheses, and verify dirty or unindexed source directly.
For structural queries, use search with action: "cypher" and repo: "<repo>".
Every Cypher query must include LIMIT of at most 5000; avoid variable-length patterns.
Use impact with maxDepth for bounded dependency traversal.
If exploration leads to edits, read ontoindex-impact-analysis before changing symbols.