Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Install and configure Cohere SDK authentication with API v2. Use when setting up a new Cohere integration, configuring API keys, or initializing the CohereClientV2 in your project. Trigger with phrases like "install cohere", "setup cohere", "cohere auth", "configure cohere API key".
.claude/skills/jeremylongshore-cohere-install-auth/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-03 | ✗→✓ | ▲ Improved | 16% | 0% |
| case-01 | ✗→✓ | ▲ Improved | 49% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-06 | ✗→✓ | ▲ Improved | 145% | 0% |
| case-12 | ✗→✓ | ▲ Improved | -19% | 0% |
Select the current v2 client for the repository, keep credentials outside source control, and prove access with a read-only Models API request.
Use Read, Glob, and Grep to inspect code, configuration, and evidence. Use WebFetch only for current Cohere primary documentation. Use Write or Edit only when the user requested implementation and the exact target files are known; never write credentials or customer content.
CohereClientV2; Python uses cohere.ClientV2.CO_API_KEY as the local injection contract or pass an approved secret reference explicitly.Use an environment-specific key injected from an approved secret manager. Never print, persist, commit, or place CO_API_KEY in an example. Confirm access with the least costly bounded operation appropriate to the task, and treat key creation, rotation, revocation, role changes, and production-capacity requests as owner-approved actions.
CO_API_KEY without writing its value, then update ignore rules if local environment files are used.Do not expose or rotate keys, change Cohere Team roles, accept commercial terms, enable sensitive production data, increase spend or capacity, switch production models, send a support bundle, or execute model-proposed side effects without the accountable owner's approval. Keep diagnosis read-only unless implementation was requested.
Return the resolved API and model contract, files or settings inspected, evidence collected, validation result, remaining risk, owner, and rollback or next action. Redact keys, authorization headers, prompts, retrieved documents, embeddings, customer identifiers, and unrestricted environment output.
| Condition | Response | |---|---| | 401 | Confirm runtime injection and key state without printing the credential. | | No models | Check team membership and deployment availability before changing code. | | SDK shape mismatch | Compare the installed major with the official generated SDK reference. | | Exposed key | Rotate or revoke it and scrub retained logs immediately. |
Use this compact handoff shape to keep the selected scope, validation evidence, and operational result reviewable.
Input:
textruntime=typescript; environment=ci; probe=models.list; secret=approved-reference
Expected handoff:
textclient=v2; sdk=cohere-ai@resolved; auth=pass; key-value=redacted
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | fail→pass | 10,354 | 6,321 | -39% | 1 | 1 | 0% | 2,115 | 2,447 | +16% | 0 | 0 | — |
case-04 | pass→pass | 6,366 | 2,368 | -63% | 1 | 1 | 0% | 1,124 | 1,587 | +41% | 0 | 0 | — |
case-01 | fail→pass | 8,771 | 6,476 | -26% | 1 | 1 | 0% | 1,686 | 2,513 | +49% | 0 | 0 | — |
case-02 | fail→pass | 9,207 | 6,508 | -29% | 1 | 1 | 0% | 1,966 | 2,459 | +25% | 0 | 0 | — |
case-05 | pass→pass | 4,825 | 3,918 | -19% | 1 | 1 | 0% | 833 | 1,801 | +116% | 0 | 0 | — |
case-06 | fail→pass | 4,729 | 4,391 | -7% | 1 | 1 | 0% | 841 | 2,064 | +145% | 0 | 0 | — |
case-07 | pass→pass | 6,662 | 3,158 | -53% | 1 | 1 | 0% | 1,295 | 1,732 | +34% | 0 | 0 | — |
case-08 | pass→pass | 6,113 | 5,346 | -13% | 1 | 1 | 0% | 1,224 | 2,071 | +69% | 0 | 0 | — |
case-09 | pass→pass | 7,792 | 2,477 | -68% | 1 | 1 | 0% | 1,438 | 1,594 | +11% | 0 | 0 | — |
case-10 | pass→pass | 5,013 | 2,577 | -49% | 1 | 1 | 0% | 892 | 1,686 | +89% | 0 | 0 | — |
case-11 | pass→pass | 3,331 | 1,498 | -55% | 1 | 1 | 0% | 674 | 1,440 | +114% | 0 | 0 | — |
case-12 | fail→pass | 11,110 | 2,723 | -75% | 1 | 1 | 0% | 2,054 | 1,672 | -19% | 0 | 0 | — |
case-13 | pass→pass | 6,371 | 1,895 | -70% | 1 | 1 | 0% | 1,228 | 1,503 | +22% | 0 | 0 | — |
case-14 | fail→pass | 9,228 | 5,516 | -40% | 1 | 1 | 0% | 1,554 | 2,179 | +40% | 0 | 0 | — |
case-15 | pass→pass | 4,172 | 1,266 | -70% | 1 | 1 | 0% | 685 | 1,259 | +84% | 0 | 0 | — |
case-16 | fail→pass | 3,624 | 995 | -73% | 1 | 1 | 0% | 613 | 1,257 | +105% | 0 | 0 | — |
case-17 | pass→pass | 9,830 | 4,242 | -57% | 1 | 1 | 0% | 1,822 | 1,969 | +8% | 0 | 0 | — |
case-18 | pass→pass | 12,770 | 7,966 | -38% | 1 | 1 | 0% | 2,358 | 2,607 | +11% | 0 | 0 | — |
case-19 | pass→pass | 7,677 | 4,618 | -40% | 1 | 1 | 0% | 1,532 | 1,961 | +28% | 0 | 0 | — |
case-20 | pass→pass | 2,452 | 1,479 | -40% | 1 | 1 | 0% | 387 | 1,413 | +265% | 0 | 0 | — |
case-21 | pass→pass | 5,049 | 2,717 | -46% | 1 | 1 | 0% | 931 | 1,663 | +79% | 0 | 0 | — |
case-22 | fail→fail | 14,672 | 15,407 | +5% | 1 | 1 | 0% | 3,243 | 4,819 | +49% | 0 | 0 | — |
case-23 | fail→fail | 15,581 | 12,773 | -18% | 1 | 1 | 0% | 3,095 | 3,764 | +22% | 0 | 0 | — |
case-24 | fail→fail | 25,032 | 23,314 | -7% | 1 | 1 | 0% | 5,229 | 6,194 | +18% | 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. The headline lift of +29 percentage points is the difference between those two pass rates over the 24 comparable cases.
The publisher has shipped newer versions since this run, so these numbers describe v1, not the version currently listed.
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.