Qdrant
Vector search and collection management on Qdrant — create/list collections, upsert points, nearest-neighbour search, retrieve, scroll, and delete. Talks directly to Qdrant's REST API and can embed plain text via OpenAI, Gemini, Cohere, or a custom OpenAI-compatible endpoint.
- SDK
- Node
- Protocol
inflowv1- Actions
- 7
- Version
- v0.1.0
- Author
- Inflowenger
- Runs on
- Any
inflowv1host - Categories
- AI and agentsDatabases and stores
Reach for it when
A vector store several services share, with filtering that narrows the search rather than the results.
The use people don't expect
It can embed the text for you — OpenAI, Gemini, Cohere, or anything OpenAI-compatible like Ollama — so a retrieval flow is two nodes, not two nodes plus an embedding service you also have to run.
Flows that use it
Each one is a shape of flow, not a template to copy. Node names are the labels as they appear on the canvas; a name with a slash means either would do.
The organisation brain, kept current by a flow
Starts withA schedule. This is a long-running flow, not a batch job.
- POSTGRES / DUCKDB
Picks up what changed since the last pass.
- QDRANT
Upserts plain text; the embedding happens in the node.
- GoTo
Loops back for the next slice. The flow survives a crash and picks up where it stopped.
What you getContext stops being a nightly export that's confidently wrong by morning. The index is maintained by a durable process you can open and read.
Retrieval scoped at the moment of the decision
Starts withA decision step inside some other flow.
- QDRANT
Nearest-neighbour search with a filter — this customer, this quarter — evaluated before the search.
- LLM / CLAUDE AGENT
Decides with that slice in front of it, not the whole corpus.
What you getThe model sees what the decision needs and nothing else, and the scoping rule is drawn on the canvas where an auditor can find it.
The reference, not the pitch
This page is about what the plugin is for. Every action, every connection field, the ports, the install one-liner and the caveats are in the catalog entry, which is the thing the maintainer updates.
Near it in the catalog
The task is "do this, here, on this machine" rather than "answer this" — and you want the work to stop and ask when it hits a decision nobody delegated.
CLICKHOUSEClickHouseMany rows, asked questions repeatedly — events, logs, metrics, anything you'll want a GROUP BY over.
CODEX AGENTCodex AgentSame shape as Claude Agent, over Codex — reach for it when you want the workspace fence enforced by the operating system rather than by the model behaving well.