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.
An exceptionally token-efficient gating skill with concrete probe criteria and hard operational limits, well organized for its size. Its main gap is that the actual integration procedure — the reason the skill exists — is delegated to an external docs URL rather than inlined.
Suggestions
Inline the minimal useAtlasChat wiring (provider setup, mount point, a short code snippet) so the skill is executable without fetching the external docs URL, which may be unavailable or version-drifted.
Add a post-integration verification step, e.g. 'Confirm the chat renders in the app shell and host-connect attempts still reject, before considering the integration done.'
Localize the docs.cognite.com guide's key steps into a references/ file so the body can stay lean while keeping the procedure on-bundle.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The 14-line body is lean with zero padding: every line is either an eligibility gate, a concrete probe ('@cognite/app-sdk + connectToHostApp / useHostApp resolving inside Fusion/EOS'), or a hard constraint ('5 per user action, ceiling 50, cache by space:externalId:lastUpdatedTime'). It assumes Claude's competence and explains nothing the model already knows. | 5 / 5 |
Actionability | Concrete guidance exists for the eligibility probe and the rate-limit/cache rules, but the primary task — 'follow https://docs.cognite.com/... and wire useAtlasChat' — contains no executable steps or code, deferring entirely to an external URL and a sibling skill. This matches 'some concrete guidance but incomplete; missing key details' rather than the mostly-executable anchor 4. | 3 / 5 |
Workflow Clarity | The numbered 1-3 gate sequence is clear, and the probe step includes explicit decision signals plus a checkpoint ('If unsure, ask. Default to the sidebar.'). It falls short of 5 because there is no validation or verification of the integration outcome (e.g. confirming the chat mounts and host-connect rejects as expected). | 4 / 5 |
Progressive Disclosure | The skill is under 50 lines with no bundle files (references/, scripts/, assets/ do not exist), and the body is organized into clear gate, instruction, and constraint sections with well-signaled pointers ('See integrate-fusion-agent', the docs URL). Per the simple-skill guideline, this warrants a 5. | 5 / 5 |
Total | 17 / 20 Passed |