Content
78%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 highly actionable, dense integration reference with executable code, precise endpoint/error semantics, and clear routing workflows. Its weaknesses are progressive disclosure (monolithic inlining of 13 playbooks plus broken example-file references) and minor conciseness redundancy.
Suggestions
Create the referenced ./examples/swap.md, lend.md, trigger.md, and price.md files (or remove the dead links), since the 'Complete Working Examples' section currently points to non-existent files.
Split the 13 inlined API playbooks into one-level-deep reference files (e.g. references/swap.md, references/trigger.md) and keep SKILL.md as a routing overview pointing to them, rather than a ~400-line monolith.
De-duplicate rate-limit guidance (it appears in both the Swap section and the Rate Limits section) and trim the single ~50-term triggers line to the highest-signal terms to recover tokens.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and assumes Claude knows Solana/web3 without over-explaining concepts, but it has minor redundancy (rate-limit guidance repeated in the Swap section and the Rate Limits section) and the ~50-term single-line triggers list is bloated, keeping it just below the lean anchor. | 4 / 5 |
Actionability | It provides copy-paste-ready TypeScript helpers (jupiterFetch, signAndSend, jupiterAction, withRetry), exact endpoint paths and base URLs, an error-code table mapping each code to a concrete retry action, and field-level gotchas (e.g. token.audit?.isSus === true). | 5 / 5 |
Workflow Clarity | The Intent Router table sequences intent to API family to first action, per-API playbooks give ordered first actions, and withRetry/jupiterAction plus the Fresh Context Policy and Production Hardening checklists provide validation and feedback loops, but the bulk of the playbooks are reference-style endpoint/gotcha lists rather than full validate-fix-retry workflows. | 4 / 5 |
Progressive Disclosure | Headers and the Intent Router give reasonable structure, but ~400 lines of inlined API reference across 13 families could be split into separate reference files, and the ./examples/swap.md|lend.md|trigger.md|price.md links are broken (no examples/ directory or any references/scripts/assets bundle exists), so references are signaled but non-functional. | 3 / 5 |
Total | 16 / 20 Passed |