Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Upload, edit, and export documents via Nudocs.ai. Use when creating shareable document links for collaborative editing, uploading markdown/docs to Nudocs for rich editing, or pulling back edited content. Triggers on "send to nudocs", "upload to nudocs", "edit in nudocs", "pull from nudocs", "get the nudocs link", "show my nudocs documents".
.claude/skills/jdrhyne-nudocs/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-04 | ✗→✓ | ▲ Improved | 175% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 111% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 27% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 94% | 0% |
| case-13 | ✗→✓ | ▲ Improved | -8% | 0% |
Use the authenticated Nudocs CLI to upload documents for hosted editing, retrieve private edit links, list or export documents, and delete a specific remote document.
Inspect nudocs --help and nudocs config --help before setup because authentication UX can change by version.
Use one authentication method, not both:
~/.config/nudocs mode 0700 and its API-key file mode 0600.NUDOCS_API_KEY only when it is already supplied through protected local or CI secret storage.Never ask the user to paste the key into chat, put it in a command argument, echo it into a file, display it with nudocs config, or print environment/config contents. Verify authentication through behavior such as a bounded nudocs list, not by inspecting the secret.
Before an operation, inspect the relevant current help:
bashnudocs --help nudocs upload --help nudocs link --help nudocs pull --help nudocs delete --help
The maintained CLI currently documents upload, list, link, pull, delete, and config. Do not invent aliases or use the obsolete gimme command.
For current upload/export formats, read references/formats.md. Do not load that reference for listing, link lookup, or deletion. Service limits are account- and time-dependent: use the CLI/API response or current Nudocs account UI instead of a hard-coded document count.
Every link, pull, and delete workflow must resolve and pass one exact document ID returned by the current nudocs list output or explicitly supplied by the user:
bashnudocs link <document-id> nudocs pull <document-id> --format <format> --output <exact-output-path> nudocs delete <document-id>
Never run bare nudocs link, nudocs pull, or nudocs delete, and never rely on a CLI default such as the last uploaded document. A title, filename, URL, phrase such as "the last upload," or other non-ID reference is not an action target. Use a bounded nudocs list to resolve it; if zero or multiple candidates remain, refuse the action and ask the user to choose an exact ID. Inspect current command help before execution and keep the resolved ID explicit in the command.
list is read-only; request only the scope needed.pull, resolve the exact document ID and output path/format, then pass that ID to nudocs pull <document-id>.Uploading sends the file to Nudocs, and the returned edit link grants access to hosted content.
An explicit request such as “upload this file to Nudocs” authorizes an ordinary non-sensitive upload after the data-boundary disclosure; it does not authorize publishing or forwarding the link.
nudocs link <document-id> retrieves the edit link for that exact ID; it does not prove that the link is public.Deletion is destructive remote state change.
nudocs delete <document-id> with the same resolved ID shown to the user.@nutrient-sdk/nudocs-cli package only with user authorization.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-15 | pass→pass | 12,770 | 18,632 | +46% | 1 | 1 | 0% | 1,147 | 1,848 | +61% | 0 | 0 | — |
case-01 | fail→fail | 13,878 | 33,951 | +145% | 1 | 1 | 0% | 585 | 1,480 | +153% | 0 | 0 | — |
case-02 | fail→fail | 8,757 | 37,315 | +326% | 1 | 1 | 0% | 348 | 1,633 | +369% | 0 | 0 | — |
case-03 | fail→fail | 13,796 | 19,864 | +44% | 1 | 1 | 0% | 1,486 | 1,616 | +9% | 0 | 0 | — |
case-04 | fail→pass | 12,610 | 13,220 | +5% | 1 | 1 | 0% | 931 | 2,556 | +175% | 0 | 0 | — |
case-05 | pass→pass | 13,039 | 11,299 | -13% | 1 | 1 | 0% | 1,380 | 1,811 | +31% | 0 | 0 | — |
case-06 | fail→pass | 27,597 | 14,476 | -48% | 1 | 1 | 0% | 1,113 | 2,347 | +111% | 0 | 0 | — |
case-07 | fail→fail | 10,894 | 20,825 | +91% | 1 | 1 | 0% | 737 | 1,783 | +142% | 0 | 0 | — |
case-08 | fail→fail | 17,451 | 20,521 | +18% | 1 | 1 | 0% | 520 | 1,553 | +199% | 0 | 0 | — |
case-09 | pass→pass | 8,468 | 8,724 | +3% | 1 | 1 | 0% | 567 | 1,709 | +201% | 0 | 0 | — |
case-10 | fail→pass | 13,846 | 13,667 | -1% | 1 | 1 | 0% | 2,012 | 2,551 | +27% | 0 | 0 | — |
case-11 | fail→pass | 9,266 | 28,649 | +209% | 1 | 1 | 0% | 1,363 | 2,649 | +94% | 0 | 0 | — |
case-12 | pass→pass | 10,791 | 2,937 | -73% | 1 | 1 | 0% | 940 | 1,637 | +74% | 0 | 0 | — |
case-13 | fail→pass | 14,709 | 8,761 | -40% | 1 | 1 | 0% | 1,881 | 1,722 | -8% | 0 | 0 | — |
case-14 | fail→fail | 13,613 | 4,703 | -65% | 1 | 1 | 0% | 1,627 | 1,665 | +2% | 0 | 0 | — |
case-16 | pass→pass | 51,651 | 32,691 | -37% | 1 | 1 | 0% | 1,635 | 2,985 | +83% | 0 | 0 | — |
case-17 | fail→pass | 17,271 | 7,640 | -56% | 1 | 1 | 0% | 838 | 1,610 | +92% | 0 | 0 | — |
case-18 | pass→pass | 4,796 | 3,702 | -23% | 1 | 1 | 0% | 760 | 1,696 | +123% | 0 | 0 | — |
case-19 | pass→pass | 6,452 | 12,787 | +98% | 1 | 1 | 0% | 958 | 2,009 | +110% | 0 | 0 | — |
case-20 | pass→pass | 24,096 | 12,272 | -49% | 1 | 1 | 0% | 940 | 2,274 | +142% | 0 | 0 | — |
case-21 | pass→pass | 9,228 | 18,744 | +103% | 1 | 1 | 0% | 522 | 1,934 | +270% | 0 | 0 | — |
case-22 | pass→pass | 15,571 | 6,056 | -61% | 1 | 1 | 0% | 1,593 | 2,219 | +39% | 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, and 17 counted toward the lift figure. The other 5 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 +27 percentage points is the difference between those two pass rates over the 17 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.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.6-flash | verified | 8/24/2026 | +23% |
Other measured skills in the registry, with their headline benchmark lift.