LlamaIndex¶
The llamaindex extra is intentionally empty while upstream
llama-index-core pulls an nltk release with no patched security path. Install
a safe LlamaIndex SDK version separately when upstream resolves that dependency.
For central Cloud Run review service deployment and Vercel AI SDK client wiring,
use the tracked deploy/agent-frameworks templates and the
Agent Framework Deploy Pack.
As Node Postprocessor¶
from director_ai.integrations.llamaindex import DirectorAIPostprocessor
postprocessor = DirectorAIPostprocessor(
facts={"capital": "Paris is the capital of France."},
threshold=0.6,
)
query_engine = index.as_query_engine(
node_postprocessors=[postprocessor],
)
response = query_engine.query("What is the capital of France?")