Content
85%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A high-quality, domain-expert body: every section carries non-obvious PrimeKG-specific caveats, the workflow is sequenced with explicit validation and error-recovery behavior, and detail is correctly split into an existing one-level-deep reference. The residual weaknesses are a commented-out get_neighbors example and a few review-hedging sentences that cost tokens without aiding execution.
Suggestions
Add one complete, runnable get_neighbors and find_paths example (with the node_type/node_source kwargs filled in from the search result) so the most common traversal calls are copy-paste ready.
Trim the review-process asides (e.g. "validation used small HTTP byte ranges and synthetic data, not the 982 MB file" and the stubbed-loader caveats) into the single Validation section or the data-contract reference, keeping the inline guidance instructional.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with non-obvious domain caveats (reverse-row symmetrization, relation semantics, namespace collisions, MONDO_grouped IDs) that Claude would not know, with no library tutorials or filler. Minor trimmable instances remain — review-hedging asides like "validation used small HTTP byte ranges and synthetic data, not the 982 MB file" and inline review dates — placing it at the efficient-with-minor-trimming anchor rather than the every-token-earns-its-place anchor. | 4 / 5 |
Actionability | Provides a copy-paste curl download, an executable search_nodes example, and exact API signatures (e.g. `get_neighbors(node_id, relation_type=None, *, node_type=None, node_source=None)`) backed by real functions in scripts/query_primekg.py. The gap to a 5 is that the get_neighbors call is commented out and find_paths lacks a full invocation example, so coverage of common cases is mostly — not fully — executable. | 4 / 5 |
Workflow Clarity | A clear numbered sequence — pin and verify the artifact, resolve a typed entity before traversal, retrieve associations, then query context/paths — with explicit validation checkpoints and feedback loops: checksum/size verification before use, reviewing candidates before calling get_neighbors, "an ambiguous name returns an error and candidates; it never silently selects the first hit", and explicit failure on unsupported depths or max_paths overflow. Queries are read-only, so the destructive-operation cap does not apply. | 5 / 5 |
Progressive Disclosure | The body is a well-organized overview that delegates data/API provenance (Dataverse metadata, checksum verification, PyTDC contracts) to a clearly signaled, one-level-deep references/data-contract.md (linked twice, file exists), and references the importable scripts.query_primekg helper (verified present). Core usage guidance (relation table, API semantics) is appropriately inline, matching the clear-overview/well-signaled-references anchor. | 5 / 5 |
Total | 18 / 20 Passed |