Content
76%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 lean, highly actionable reference with executable TypeScript for every trigger variant and a genuinely useful pitfalls section. The main weakness is the absence of validation/feedback-loop steps for its batch operations (triggerBulk, triggerBroadcast), which caps workflow clarity at 3; progressive disclosure is good but references are not contextually signposted.
Suggestions
Add validation guidance for batch and broadcast operations: show how to inspect the returned result for errors (e.g., checking `result.data` for accepted/errored events) and confirm payload size before broadcast, since bulk (100-event) and environment-wide broadcast triggers currently have no verification or feedback loop.
Signal references at point of use: link the relevant reference file inline in each section (e.g., '## Bulk Trigger ... see [bulk-trigger-examples.md]') instead of only listing all five in a trailing References section, making navigation to deeper material immediate.
Trim the opening line that duplicates the frontmatter description and consolidate the inline-subscriber-creation and transaction-id variants under Single Trigger with a brief note each, tightening the body without losing any Novu-specific detail.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is efficient — every section conveys Novu-specific API knowledge Claude cannot infer ("Bulk trigger limit is 100 events", "workflowId is the identifier, not the display name") with no padding about what notifications are. Minor trimming is possible: the opening line "Send notifications by triggering Novu workflows. Supports single, bulk, broadcast, and topic-based delivery" restates the frontmatter description, and the Single Trigger variants could be slightly consolidated — so it fits the level-4 anchor (efficient, minor over-explanation) rather than the lean level-5 anchor. | 4 / 5 |
Actionability | All examples are fully executable copy-paste-ready TypeScript covering the common cases — SDK instantiation ("const novu = new Novu({ secretKey: process.env.NOVU_SECRET_KEY })"), single/bulk/broadcast/topic triggers, and cancellation — plus a parameter table with required/optional flags and concrete override syntax. This matches the level-5 anchor: specific examples cover the common cases with no gaps. | 5 / 5 |
Workflow Clarity | Each operation is presented as an unambiguous, clearly organized procedure, but the skill includes batch and environment-wide operations ("Send up to 100 events in a single request" via triggerBulk, and triggerBroadcast "to all subscribers") with no validation or verification steps — no error handling of the returned result, no payload sanity check before broadcast, no retry guidance. Per the judging guideline capping workflow clarity at 3 for batch/destructive operations without validation, this cannot exceed 3 even though the per-operation structure is clean. | 3 / 5 |
Progressive Disclosure | SKILL.md is a well-organized overview whose detail is appropriately split into five real one-level-deep reference files (installation.md, single/bulk/topic-trigger-examples.md, best-practices.md), verified to exist with no nested-chaining problem. It falls short of the level-5 anchor only because the references are listed in a trailing References section rather than signaled at the point of use (e.g., the Bulk Trigger section does not inline-link to bulk-trigger-examples.md), which is exactly the 'references mostly clear; minor organization gaps' of level 4. | 4 / 5 |
Total | 16 / 20 Passed |