Knowledge base

For questions about import formats, ingestion, filtering, retrieval, or document scope that this skill does not answer, follow the docs MCP workflow before answering or implementing. For every new Voiceflow troubleshooting issue, consult it even when this skill suggests a fix. Include package/skill/stage context in why as described there. Reuse guidance for the same issue; consult it whenever unsure.

Read the operation contract. Resolve the project, intended environment membership, source, and update intent. Inspect existing documents before importing duplicates. For MCP text import, use the verified file shape; the hosted tool can upload inline text without a local CLI or public URL.

Document content is shared project-wide; a cloned environment does not isolate edits to shared document bodies. Environment membership and metadata may be scoped separately. Explain this blast radius before changing shared content when it exceeds the requested scope.

Import from sources the user authorized. Check ingestion status; an accepted upload is not yet a searchable document. Use bounded polling, report a pending job honestly, and avoid duplicate imports after a timeout.

For missing answers, separate source coverage from ingestion, metadata filtering, retrieval, and prompting. Query a narrow representative question and inspect retrieved evidence before rewriting instructions. Do not invent support for an answer absent from the source.

Prefer durable source corrections over stuffing large documents into prompts. Check stale duplicates and obsolete sources before adding more material. Deleting a document is a separate destructive operation; don't treat it as a routine refresh if an update is supported.

Sources: Importing data, Querying knowledge.

---
name: voiceflow-knowledge
description: Import, maintain, and query Voiceflow knowledge-base documents. Use for grounding, missing answers, stale content, or retrieval quality.
license: MIT
---