AI repos pipeline (ArcPX-style)

File: examples/pipeline-ai-repos.normalized.json
Derived from a real ArcPX export. Requires a custom nodeExecutor (and BYO LLM for genText).

Placeholder chain ({$aiRepos} → pipe → {$adaptive_json_agent} → genText):

Nodes

aiRepos (data source)

data.nodeData is an object:

{
  "company": ["nvidia", "openai", "anthropic", "google-deepmind", "meta", "huggingface"],
  "limit": 2,
  "fetch_readme": false
}

pipeNode (transform)

• outputTarget: "$adaptive_json_agent"
• nodeData: prompt template with {$aiRepos} placeholder

genText (LLM)

• isPassThrough: true
• nodeData: prompt with {$adaptive_json_agent}
• Receives aggregated sourceData at run time

Runtime

The engine augments each node with data.sourceData and data.settings before nodeExecutor runs. Implement {$variable} substitution and API/LLM calls in your handler.

Payload guide · Docs index