Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Import Vapi voice assistants into Telnyx with all configurations — instructions, greeting, voice settings, tools, and call analysis. Supports selective import by assistant ID and covers all SDK languages.
.claude/skills/team-telnyx-telnyx-import-vapi/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 11% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 25% | 0% |
| case-03 | ✗→✓ | ▲ Improved | 15% | 0% |
| case-04 | ✗→✓ | ▲ Improved | 4% | 0% |
| case-05 | ✗→✓ | ▲ Improved | 15% | 0% |
Migrate your Vapi voice assistants to Telnyx in minutes. The import API pulls assistant configurations directly from Vapi using your API key and recreates them as Telnyx AI Assistants.
Interaction model: Collect the user's Telnyx API key and Vapi API key, store the Vapi key as a Telnyx integration secret, run the import, then verify. Do NOT skip the secret-creation step — the import endpoint requires a secret reference, not a raw key.
| Component | Imported? | Notes | |-----------|-----------|-------| | Instructions | Yes | Imported as-is | | Greeting / first message | Yes | Maps to assistant greeting | | Voice configuration | Yes | Voice provider and voice ID preserved | | Dynamic variables | Yes | Default values carried over | | Tools (hangup, transfer, webhook) | Yes | Tool definitions and configurations | | MCP Server integrations | Yes | Server URLs and tool mappings | | Call analysis / insights | Yes | Mapped to insight_settings | | Data retention preferences | Yes | Mapped to privacy_settings | | Knowledge base | No | Must be manually added post-import | | Secrets (API keys in tools) | Partial | Placeholder secrets created — you must re-enter values in the Telnyx portal |
Before importing, store your Vapi API key as an integration secret in Telnyx. Note the secret reference name (e.g., vapi_api_key) — you'll use it in the import call.
You can create integration secrets via the Telnyx Portal under Integration Secrets, or via the API.
Import every assistant from your Vapi account:
bashcurl \ -X POST \ -H "Authorization: Bearer $TELNYX_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "provider": "vapi", "api_key_ref": "vapi_api_key" }' \ "https://api.telnyx.com/v2/ai/assistants/import"
pythonimport os from telnyx import Telnyx client = Telnyx(api_key=os.environ.get("TELNYX_API_KEY")) assistants = client.ai.assistants.imports( provider="vapi", api_key_ref="vapi_api_key", ) for assistant in assistants.data: print(f"Imported: {assistant.name} (ID: {assistant.id})")
javascriptimport Telnyx from 'telnyx'; const client = new Telnyx(); const assistants = await client.ai.assistants.imports({ provider: 'vapi', api_key_ref: 'vapi_api_key', }); for (const assistant of assistants.data) { console.log(`Imported: ${assistant.name} (ID: ${assistant.id})`); }
goassistants, err := client.AI.Assistants.Imports(context.TODO(), telnyx.AIAssistantImportsParams{ Provider: telnyx.AIAssistantImportsParamsProviderVapi, APIKeyRef: "vapi_api_key", }) if err != nil { panic(err.Error()) } for _, a := range assistants.Data { fmt.Printf("Imported: %s (ID: %s)\n", a.Name, a.ID) }
javaimport com.telnyx.sdk.models.ai.assistants.AssistantImportsParams; import com.telnyx.sdk.models.ai.assistants.AssistantsList; AssistantImportsParams params = AssistantImportsParams.builder() .provider(AssistantImportsParams.Provider.VAPI) .apiKeyRef("vapi_api_key") .build(); AssistantsList assistants = client.ai().assistants().imports(params); assistants.getData().forEach(a -> System.out.printf("Imported: %s (ID: %s)%n", a.getName(), a.getId()));
rubyassistants = client.ai.assistants.imports( provider: :vapi, api_key_ref: "vapi_api_key" ) assistants.data.each do |a| puts "Imported: #{a.name} (ID: #{a.id})" end
To import only certain assistants, pass their Vapi assistant IDs in import_ids:
bashcurl \ -X POST \ -H "Authorization: Bearer $TELNYX_API_KEY" \ -H "Content-Type: application/json" \ -d '{ "provider": "vapi", "api_key_ref": "vapi_api_key", "import_ids": ["vapi-assistant-id-1", "vapi-assistant-id-2"] }' \ "https://api.telnyx.com/v2/ai/assistants/import"
pythonassistants = client.ai.assistants.imports( provider="vapi", api_key_ref="vapi_api_key", import_ids=["vapi-assistant-id-1", "vapi-assistant-id-2"], )
javascriptconst assistants = await client.ai.assistants.imports({ provider: 'vapi', api_key_ref: 'vapi_api_key', import_ids: ['vapi-assistant-id-1', 'vapi-assistant-id-2'], });
List your Telnyx assistants to confirm the import succeeded:
bashcurl -H "Authorization: Bearer $TELNYX_API_KEY" \ "https://api.telnyx.com/v2/ai/assistants"
pythonassistants = client.ai.assistants.list() for a in assistants.data: print(f"{a.name} — {a.id} — imported: {a.import_metadata}")
javascriptconst assistants = await client.ai.assistants.list(); for (const a of assistants.data) { console.log(`${a.name} — ${a.id} — imported:`, a.import_metadata); }
After importing, complete these manual steps:
Running the import again for the same Vapi assistants will overwrite the existing Telnyx copies with the latest configuration from Vapi. This is useful for syncing changes during a gradual migration.
| Field | Type | Required | Description | |-------|------|----------|-------------| | provider | string | Yes | Must be "vapi" | | api_key_ref | string | Yes | Name of the Telnyx integration secret containing your Vapi API key | | import_ids | arraystring] | No | Specific Vapi assistant IDs to import. Omit to import all. |
Endpoint: POST https://api.telnyx.com/v2/ai/assistants/import
Full API docs: https://developers.telnyx.com/api-reference/assistants/import-assistants-from-external-provider
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 19,754 | 11,202 | -43% | 1 | 1 | 0% | 3,753 | 4,177 | +11% | 0 | 0 | — |
case-02 | fail→pass | 16,191 | 9,107 | -44% | 1 | 1 | 0% | 3,056 | 3,808 | +25% | 0 | 0 | — |
case-03 | fail→pass | 20,728 | 10,589 | -49% | 1 | 1 | 0% | 3,584 | 4,110 | +15% | 0 | 0 | — |
case-22 | pass→pass | 11,759 | 13,592 | +16% | 1 | 1 | 0% | 2,167 | 4,436 | +105% | 0 | 0 | — |
case-21 | pass→fail | 6,613 | 5,008 | -24% | 1 | 1 | 0% | 1,287 | 3,030 | +135% | 0 | 0 | — |
case-04 | fail→pass | 15,964 | 5,910 | -63% | 1 | 1 | 0% | 2,984 | 3,113 | +4% | 0 | 0 | — |
case-05 | fail→pass | 18,236 | 8,163 | -55% | 1 | 1 | 0% | 3,224 | 3,703 | +15% | 0 | 0 | — |
case-06 | fail→pass | 12,992 | 7,979 | -39% | 1 | 1 | 0% | 2,692 | 3,484 | +29% | 0 | 0 | — |
case-07 | fail→pass | 10,617 | 6,625 | -38% | 1 | 1 | 0% | 1,760 | 3,263 | +85% | 0 | 0 | — |
case-08 | pass→pass | 12,055 | 2,906 | -76% | 1 | 1 | 0% | 1,819 | 2,426 | +33% | 0 | 0 | — |
case-09 | fail→pass | 7,865 | 2,621 | -67% | 1 | 1 | 0% | 1,470 | 2,521 | +71% | 0 | 0 | — |
case-10 | pass→pass | 11,177 | 4,281 | -62% | 1 | 1 | 0% | 1,884 | 2,777 | +47% | 0 | 0 | — |
case-11 | fail→pass | 12,963 | 3,439 | -73% | 1 | 1 | 0% | 1,975 | 2,496 | +26% | 0 | 0 | — |
case-12 | fail→pass | 16,209 | 3,292 | -80% | 1 | 1 | 0% | 2,495 | 2,562 | +3% | 0 | 0 | — |
case-13 | fail→pass | 12,098 | 2,947 | -76% | 1 | 1 | 0% | 1,900 | 2,490 | +31% | 0 | 0 | — |
case-14 | pass→pass | 11,377 | 6,459 | -43% | 1 | 1 | 0% | 1,839 | 3,116 | +69% | 0 | 0 | — |
case-15 | fail→pass | 11,034 | 3,313 | -70% | 1 | 1 | 0% | 2,103 | 2,603 | +24% | 0 | 0 | — |
case-16 | fail→pass | 9,020 | 4,818 | -47% | 1 | 1 | 0% | 1,727 | 2,958 | +71% | 0 | 0 | — |
case-17 | pass→pass | 10,039 | 2,524 | -75% | 1 | 1 | 0% | 2,058 | 2,397 | +16% | 0 | 0 | — |
case-18 | fail→pass | 6,621 | 2,550 | -61% | 1 | 1 | 0% | 1,240 | 2,446 | +97% | 0 | 0 | — |
case-19 | fail→pass | 14,703 | 4,093 | -72% | 1 | 1 | 0% | 2,246 | 2,676 | +19% | 0 | 0 | — |
case-20 | pass→pass | 15,589 | 12,078 | -23% | 1 | 1 | 0% | 2,924 | 4,338 | +48% | 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 +64 percentage points is the difference between those two pass rates over the 22 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.