Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →nwave-ai/nw-por-review-criteria
Review dimensions and bug patterns for journey artifact reviews
Passed 25 of 25 test cases with the skill, and 7 of 25 without it.
and 24 more in the full report
19 fixed7 passed either way0 still failing0 broke
Tokens, turns and time describe resource use, not a dollar price.
One of 18 test cases the skill fixed. Same model, same prompt, with and without the skill.
List the exact top-level YAML metadata keys required to identify the review session, the reviewer agent, and the target file in an Eclipse product owner review document.
| Checked | Without skill | With skill |
|---|---|---|
| Includes top-level YAML keys review_id, reviewer, and artifact_reviewed. | Missed | Passed |
The instructions your model reads when the skill loads.
Domain knowledge for product-owner-reviewer (Eclipse). Covers journey coherence, emotional arcs, shared artifacts, example data quality, CLI UX patterns.
Validate complete flow with no gaps.
Checks: all steps start-to-goal defined | no orphan steps | no dead ends | decision branches lead somewhere | error paths guide to recovery
Severity: critical = missing main flow steps / dead ends | high = orphan steps | medium = ambiguous decisions | low = minor clarity
Validate emotional design quality.
Checks: arc defined (start/middle/end) | all steps annotated | no jarring transitions | confidence builds progressively | error states guide not frustrate
Severity: critical = no arc / major jarring transitions | high = missing key annotations | medium = confidence doesn't build | low = minor polish
Validate ${variable} sources and consistency.
Checks: all ${variables} have documented source | single source of truth | all consumers listed | integration risks assessed | validation methods specified
Severity: critical = undocumented ${variables} / multiple sources | high = missing consumers / unassessed risks | medium = incomplete validation | low = minor consumer docs
Key review skill -- analyze data for integration gaps.
Checks: realistic not generic | reveals integration dependencies | catches version mismatches | catches path inconsistencies | consistent across steps
Severity: critical = generic placeholders hide issues | high = inconsistent across steps | medium = doesn't reveal deps | low = could be more realistic
Apply: 1) trace ${version} through all steps -- same? 2) compare ${install_path} step 2 vs 3 -- match? 3) does data show actual integration points?
Generic "v1.0.0" or "/path/to/install" hides bugs. Realistic "v1.2.86" from "pyproject.toml" reveals bugs.
Checks: command vocabulary consistent | help available | error messages guide to resolution | progressive disclosure respected
test cases passed, vs 7 / 25 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.