Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Bulk extract content from a website. Crawls pages following links up to a depth/limit.
.claude/skills/firecrawl-crawl/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 21% | 0% |
| case-01 | ✗→✓ | ▲ Improved | -69% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 122% | 0% |
| case-04 | ✗→✓ | ▲ Improved | -40% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -41% | 0% |
Bulk extract content from a website. Crawls pages following links up to a depth/limit.
Prerequisite: crawl requires authentication (no keyless free tier); without credentials the CLI prompts an interactive login.
bash# Crawl a docs section firecrawl crawl "<url>" --include-paths /docs --limit 50 --wait -o .firecrawl/crawl.json # Full crawl with depth limit firecrawl crawl "<url>" --max-depth 3 --wait --progress -o .firecrawl/crawl.json # Check status of a running crawl firecrawl crawl <job-id>
Run firecrawl crawl --help for the full option list.
Done when: the crawl reaches a terminal status and the saved output under .firecrawl/ contains the expected pages.
--wait when you need the results immediately. It has no default timeout; use --timeout <seconds> to bound polling. Without --wait, crawl returns a job ID for async polling.--include-paths whenever the request names a section — crawl only the pages you need.firecrawl credit-usage before large crawls (credit-usage requires authentication).| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | fail→pass | 9,699 | 13,537 | +40% | 1 | 1 | 0% | 555 | 672 | +21% | 0 | 0 | — |
case-01 | fail→pass | 24,472 | 8,486 | -65% | 1 | 1 | 0% | 3,027 | 943 | -69% | 0 | 0 | — |
case-02 | fail→pass | 29,843 | 6,547 | -78% | 1 | 1 | 0% | 282 | 625 | +122% | 0 | 0 | — |
case-04 | fail→pass | 14,715 | 8,736 | -41% | 1 | 1 | 0% | 1,459 | 880 | -40% | 0 | 0 | — |
case-05 | pass→pass | 16,461 | 8,281 | -50% | 1 | 1 | 0% | 2,067 | 932 | -55% | 0 | 0 | — |
case-06 | fail→pass | 15,477 | 8,067 | -48% | 1 | 1 | 0% | 1,344 | 794 | -41% | 0 | 0 | — |
case-07 | fail→pass | 20,748 | 7,277 | -65% | 1 | 1 | 0% | 2,559 | 751 | -71% | 0 | 0 | — |
case-08 | fail→pass | 20,649 | 7,739 | -63% | 1 | 1 | 0% | 2,431 | 784 | -68% | 0 | 0 | — |
case-09 | fail→pass | 9,942 | 7,817 | -21% | 1 | 1 | 0% | 943 | 920 | -2% | 0 | 0 | — |
case-10 | fail→pass | 12,330 | 10,379 | -16% | 1 | 1 | 0% | 2,024 | 1,125 | -44% | 0 | 0 | — |
case-11 | fail→fail | 14,731 | 7,316 | -50% | 1 | 1 | 0% | 1,835 | 702 | -62% | 0 | 0 | — |
case-12 | fail→pass | 10,958 | 8,139 | -26% | 1 | 1 | 0% | 881 | 948 | +8% | 0 | 0 | — |
case-13 | fail→pass | 34,484 | 1,933 | -94% | 1 | 1 | 0% | 5,522 | 721 | -87% | 0 | 0 | — |
case-14 | fail→pass | 17,885 | 10,991 | -39% | 1 | 1 | 0% | 1,764 | 1,290 | -27% | 0 | 0 | — |
case-15 | fail→pass | 19,356 | 3,782 | -80% | 1 | 1 | 0% | 2,085 | 851 | -59% | 0 | 0 | — |
case-16 | fail→fail | 17,527 | 6,541 | -63% | 1 | 1 | 0% | 1,665 | 606 | -64% | 0 | 0 | — |
case-17 | pass→pass | 16,507 | 3,049 | -82% | 1 | 1 | 0% | 1,176 | 762 | -35% | 0 | 0 | — |
case-18 | fail→pass | 22,600 | 2,471 | -89% | 1 | 1 | 0% | 3,355 | 682 | -80% | 0 | 0 | — |
case-19 | fail→pass | 21,627 | 8,028 | -63% | 1 | 1 | 0% | 3,165 | 891 | -72% | 0 | 0 | — |
case-20 | fail→pass | 9,628 | 4,393 | -54% | 1 | 1 | 0% | 1,803 | 1,190 | -34% | 0 | 0 | — |
case-21 | fail→pass | 17,568 | 9,744 | -45% | 1 | 1 | 0% | 1,915 | 978 | -49% | 0 | 0 | — |
case-22 | pass→pass | 19,610 | 19,608 | -0% | 1 | 1 | 0% | 2,360 | 1,822 | -23% | 0 | 0 | — |
case-23 | fail→pass | 17,373 | 5,621 | -68% | 1 | 1 | 0% | 2,087 | 1,417 | -32% | 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. 23 cases were attempted, and 21 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +78 percentage points is the difference between those two pass rates over the 21 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/23/2026 | +70% |
| gemini-3.6-flash | verified | 7/23/2026 | +68% |
Other measured skills in the registry, with their headline benchmark lift.