Usage overview¶
Both nodes are in the AI Context category of the FlowDrop editor palette.
| Node | Use it when | Plugin ID |
|---|---|---|
| AI Context Item | You know which item the agent needs, every time. | ai_context_injector |
| AI Context | The right items depend on the task the workflow receives. | ai_context_dynamic_selector |
The rendered block¶
Both nodes produce rendered_text: the structured block (id, label, purpose,
guidance) that AI Context builds when it injects context into a system prompt.
The agent receives context in the same shape whether a workflow or AI Context
supplied it.
Connect rendered_text to additional_context on the AI Agent Executor node.
flowchart LR
Trigger[Trigger / input] -- message --> Sel[AI Context]
Trigger -- message --> Agent[AI Agent Executor]
Sel -- rendered_text --> Agent
Success and errors¶
Neither node stops the workflow when it cannot load context. Both return
success and error_message, with the other outputs empty on failure. Route
on success if the agent must not run without context.
An empty selection on the AI Context node is not an error: success is TRUE
and item_count is 0.
Usage tracking¶
When AI Context usage tracking is enabled, each run records which items were injected. The runner ID is the FlowDrop pipeline ID, or the workflow ID when there is no pipeline. Usage is attributed to:
- the agent (
agent:<id>) when the AI Context node inherits an agent's configuration; - otherwise the workflow node, as
flowdrop:<workflow id>:<node id>.
When the AI Context node has a target entity, the entity is attached to the usage record.