Voiceflow

@Voiceflow export the production environment configuration so I can review all workflows and variables before we migrate to the new project structure.

Exported production environment snapshot (3.2 MB JSON). The export includes 47 workflows, 12 custom functions, 23 user-defined variables, 8 API integrations, and 156 knowledge-base documents. All agent prompts, intents, and conversation paths are preserved in the snapshot. You can now diff this against your staging environment or use it as a rollback reference.

Export captures a complete point-in-time snapshot of your Voiceflow environment. Useful for version control, audits, or migrations. Requires API key with read access to the target environment. Large projects may produce multi-megabyte JSON files.

@Voiceflow create a variable called 'user_tier' with default value 'free' in the staging environment so we can start testing subscription-based routing logic.

Created variable 'user_tier' (ID: var_8x3k9m2p) in staging environment. Default value set to 'free'. The variable is now available in all workflows and can be referenced in conditions, API calls, or agent responses. Remember to update your conversation flows to populate this variable based on user authentication data.

Variable creation is immediate but requires the API key to have write permissions on the target environment. Once created, the variable persists across all workflows in that environment. Deleting a variable later requires confirming it's not referenced in active flows to avoid runtime errors.

@Voiceflow pull the last 15 transcripts from our support bot and summarize the most common user complaints so we can prioritize fixes this sprint.

Retrieved 15 recent transcripts (spanning 4 days). Common complaint themes: (1) 'Refund policy unclear' — mentioned in 8 conversations, users repeatedly asked follow-up questions. (2) 'Can't find order status' — 5 users abandoned after the bot failed to parse order numbers with dashes. (3) 'Transfer to human agent' — 6 users requested escalation, average wait before request was 2.3 turns. Recommend clarifying refund copy in the knowledge base and adding order-number format validation.

Transcript retrieval gives you raw conversation logs, user inputs, bot responses, and metadata like duration and evaluation scores. Pairing this with AI summarization helps surface patterns across sessions. Transcripts are only available for environments where logging is enabled, and retention policies vary by Voiceflow plan tier.

When Voiceflow MCP helps debug production conversation flows

A 6-person conversational AI team ships a customer-support voice agent and needs to triage why callers drop off at the billing step. The Voiceflow MCP wins here: Get Transcript pulls the full conversation log with execution stack and variables, then Get Conversation State shows exactly which branch the user hit before hanging up. The team spots a missing variable handoff in 10 minutes instead of guessing from analytics dashboards. This works when you own the Voiceflow project and have API-key access to the workspace. If you're debugging a third-party agent or lack workspace admin rights, you can't authenticate and this MCP is a non-starter. For teams shipping voice or chat agents on Voiceflow who need fast root-cause analysis during on-call shifts, this is the right call.

Automating help-doc ingestion into Voiceflow knowledge bases

A 3-person product team runs a Voiceflow chatbot that answers SaaS onboarding questions, and the help docs live in Notion. Every sprint they publish 4-6 new articles and want the bot to stay current without manual copy-paste. The Import Knowledge Document from URL tool crawls each published URL and adds it to the knowledge base in one call. This scenario works when your docs are public or behind a URL the MCP can reach, and when you're okay with quota consumption per import. If your help center is gated or you need to dedupe before ingesting, you'll hit friction. For small teams that ship docs faster than they can update the bot manually, this MCP closes the loop and keeps the agent accurate without a dedicated ops person.

When to use Voiceflow MCP for staging-to-prod promotion

A 5-person enterprise chatbot team maintains separate Voiceflow environments for dev, staging, and production. After QA signs off on staging, they want to promote the agent config and variables to prod without clicking through the UI or risking manual drift. Export Environment pulls the full JSON snapshot from staging, then Create Variable and Get Agent Configuration let them script the promotion with a diff check. This works when your environments live in the same Voiceflow workspace and you have API-key access to both. If environments span workspaces or you need approval gates that Voiceflow's native publish flow already handles, the MCP adds complexity without payoff. For teams running formal release cycles with multiple environments, this MCP turns promotion into a repeatable script instead of a risky manual checklist.