Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when reviewing templates, rendered HTML, or shared components related to Add Subresource Integrity to external scripts. Validate the final browser-facing markup, not just the source framework abstraction.
.claude/skills/thedaviddias-subresource-integrity/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-21 | ✗→✓ | ▲ Improved | -35% | 0% |
| case-09 | ✓→✓ | ＝ Same ✓ | 11% | 0% |
| case-14 | ✓→✓ | ＝ Same ✓ | -19% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | -15% | 0% |
| case-03 | ✓→✓ | ＝ Same ✓ | 9% | 0% |
When you load JavaScript from a CDN, you're trusting that CDN completely — if it's compromised, attackers can serve malicious JavaScript to all your users. SRI adds a cryptographic hash to the tag; the browser refuses to execute the script if the hash doesn't match the downloaded content, protecting users even if the CDN is compromised or the URL is hijacked.
Find all external <script> and <link rel=stylesheet> tags in this HTML that load from CDNs and don't have integrity attributes. List each one.
Add appropriate integrity and crossorigin attributes to external CDN resources. Generate the SHA-384 hashes for each resource.
Explain Subresource Integrity, how it protects against CDN compromise, how to generate SRI hashes, and its limitations.
Review templates, server-rendered HTML, and shared components that output markup related to Add Subresource Integrity to external scripts. Flag exact elements, attributes, and routes where the rendered HTML violates the rule.
For full implementation details, code examples, and framework-specific guidance, see references/rule.md.
Rule page: https://frontendchecklist.io/en/rules/html/subresource-integrity
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-09 | pass→pass | 12,700 | 14,094 | +11% | 1 | 1 | 0% | 2,306 | 2,569 | +11% | 0 | 0 | — |
case-14 | pass→pass | 12,636 | 9,007 | -29% | 1 | 1 | 0% | 2,178 | 1,765 | -19% | 0 | 0 | — |
case-01 | fail→fail | 14,728 | 9,502 | -35% | 1 | 1 | 0% | 2,697 | 2,018 | -25% | 0 | 0 | — |
case-02 | pass→pass | 8,275 | 5,177 | -37% | 1 | 1 | 0% | 1,610 | 1,361 | -15% | 0 | 0 | — |
case-03 | pass→pass | 4,199 | 2,300 | -45% | 1 | 1 | 0% | 680 | 742 | +9% | 0 | 0 | — |
case-04 | pass→pass | 4,530 | 5,336 | +18% | 1 | 1 | 0% | 861 | 1,374 | +60% | 0 | 0 | — |
case-05 | pass→pass | 13,266 | 12,892 | -3% | 1 | 1 | 0% | 1,353 | 1,617 | +20% | 0 | 0 | — |
case-06 | pass→pass | 4,491 | 8,564 | +91% | 1 | 1 | 0% | 887 | 2,004 | +126% | 0 | 0 | — |
case-07 | pass→pass | 10,589 | 5,402 | -49% | 1 | 1 | 0% | 1,716 | 1,299 | -24% | 0 | 0 | — |
case-08 | pass→pass | 11,198 | 7,838 | -30% | 1 | 1 | 0% | 2,050 | 1,753 | -14% | 0 | 0 | — |
case-10 | pass→pass | 9,774 | 7,202 | -26% | 1 | 1 | 0% | 1,837 | 1,748 | -5% | 0 | 0 | — |
case-11 | pass→pass | 10,520 | 5,474 | -48% | 1 | 1 | 0% | 1,783 | 1,260 | -29% | 0 | 0 | — |
case-12 | pass→pass | 5,424 | 7,905 | +46% | 1 | 1 | 0% | 1,117 | 1,834 | +64% | 0 | 0 | — |
case-13 | pass→pass | 18,151 | 12,036 | -34% | 1 | 1 | 0% | 1,587 | 2,588 | +63% | 0 | 0 | — |
case-15 | pass→pass | 8,918 | 4,304 | -52% | 1 | 1 | 0% | 1,357 | 1,091 | -20% | 0 | 0 | — |
case-16 | pass→pass | 6,660 | 5,909 | -11% | 1 | 1 | 0% | 1,151 | 1,317 | +14% | 0 | 0 | — |
case-17 | pass→pass | 6,139 | 5,358 | -13% | 1 | 1 | 0% | 1,197 | 1,250 | +4% | 0 | 0 | — |
case-18 | pass→pass | 6,191 | 4,038 | -35% | 1 | 1 | 0% | 1,012 | 904 | -11% | 0 | 0 | — |
case-19 | pass→pass | 13,087 | 8,336 | -36% | 1 | 1 | 0% | 2,244 | 1,770 | -21% | 0 | 0 | — |
case-20 | fail→fail | 6,614 | 8,219 | +24% | 1 | 1 | 0% | 1,219 | 1,877 | +54% | 0 | 0 | — |
case-21 | fail→pass | 10,358 | 5,178 | -50% | 1 | 1 | 0% | 1,906 | 1,247 | -35% | 0 | 0 | — |
case-22 | pass→pass | 6,459 | 5,549 | -14% | 1 | 1 | 0% | 1,039 | 1,189 | +14% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +5 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.