Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Access DingTalk documents and knowledge spaces through the DingTalk document MCP. Use this skill whenever the user selects a DingTalk knowledge space, folder, or document.
.claude/skills/wecode-ai-dingtalk-documents/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | -34% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -39% | 0% |
| case-06 | ✗→✓ | ▲ Improved | -53% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -42% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -54% | 0% |
Use the DingTalk document MCP directly. Resource IDs in <selected_knowledge_sources> are DingTalk-native IDs.
list_nodes with its workspace ID and traverse the returned hierarchy.list_nodes with the selected folder ID and continue through descendants when needed.get_document_content with the selected node ID to read its Markdown body.search_documents does not guarantee workspace or folder scoping. Do not use an unscoped search when the user selected a narrower range; traverse the hierarchy or read the exact document instead.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→fail | 5,657 | 7,049 | +25% | 1 | 1 | 0% | 656 | 472 | -28% | 0 | 0 | — |
case-02 | fail→fail | 5,928 | 8,430 | +42% | 1 | 1 | 0% | 833 | 583 | -30% | 0 | 0 | — |
case-03 | fail→fail | 3,107 | 6,213 | +100% | 1 | 1 | 0% | 400 | 434 | +9% | 0 | 0 | — |
case-04 | fail→pass | 9,555 | 5,187 | -46% | 1 | 1 | 0% | 1,425 | 945 | -34% | 0 | 0 | — |
case-05 | fail→pass | 8,464 | 3,684 | -56% | 1 | 1 | 0% | 1,159 | 702 | -39% | 0 | 0 | — |
case-06 | fail→pass | 8,650 | 2,920 | -66% | 1 | 1 | 0% | 1,178 | 554 | -53% | 0 | 0 | — |
case-07 | pass→pass | 7,454 | 4,303 | -42% | 1 | 1 | 0% | 1,116 | 586 | -47% | 0 | 0 | — |
case-08 | fail→pass | 12,222 | 5,897 | -52% | 1 | 1 | 0% | 1,987 | 1,160 | -42% | 0 | 0 | — |
case-09 | pass→pass | 18,243 | 6,758 | -63% | 1 | 1 | 0% | 2,475 | 1,155 | -53% | 0 | 0 | — |
case-10 | pass→pass | 9,872 | 4,541 | -54% | 1 | 1 | 0% | 1,443 | 743 | -49% | 0 | 0 | — |
case-11 | pass→pass | 12,790 | 4,416 | -65% | 1 | 1 | 0% | 2,016 | 691 | -66% | 0 | 0 | — |
case-12 | pass→pass | 13,439 | 4,772 | -64% | 1 | 1 | 0% | 2,084 | 583 | -72% | 0 | 0 | — |
case-13 | fail→pass | 10,966 | 3,912 | -64% | 1 | 1 | 0% | 1,630 | 746 | -54% | 0 | 0 | — |
case-14 | pass→pass | 11,057 | 2,863 | -74% | 1 | 1 | 0% | 1,659 | 450 | -73% | 0 | 0 | — |
case-15 | fail→pass | 5,291 | 3,059 | -42% | 1 | 1 | 0% | 666 | 566 | -15% | 0 | 0 | — |
case-16 | fail→pass | 6,752 | 3,284 | -51% | 1 | 1 | 0% | 891 | 564 | -37% | 0 | 0 | — |
case-17 | fail→pass | 18,121 | 7,327 | -60% | 1 | 1 | 0% | 2,657 | 1,273 | -52% | 0 | 0 | — |
case-18 | fail→pass | 7,803 | 5,697 | -27% | 1 | 1 | 0% | 1,121 | 1,078 | -4% | 0 | 0 | — |
case-19 | pass→pass | 16,731 | 2,878 | -83% | 1 | 1 | 0% | 2,427 | 559 | -77% | 0 | 0 | — |
case-20 | pass→fail | 11,215 | 7,575 | -32% | 1 | 1 | 0% | 1,892 | 577 | -70% | 0 | 0 | — |
case-21 | fail→fail | 10,112 | 12,282 | +21% | 1 | 1 | 0% | 1,725 | 1,115 | -35% | 0 | 0 | — |
case-22 | pass→pass | 6,940 | 7,124 | +3% | 1 | 1 | 0% | 1,205 | 1,296 | +8% | 0 | 0 | — |
case-23 | fail→pass | 5,383 | 3,409 | -37% | 1 | 1 | 0% | 737 | 658 | -11% | 0 | 0 | — |
case-24 | pass→pass | 11,404 | 2,307 | -80% | 1 | 1 | 0% | 1,645 | 475 | -71% | 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. 24 cases were attempted, and 20 counted toward the lift figure. The other 4 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 +38 percentage points is the difference between those two pass rates over the 20 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
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.