Claus Ibsen created CAMEL-23860:
-----------------------------------
Summary: camel-langchain4j-chat/agent/tools - Add token usage and
finish reason as exchange headers
Key: CAMEL-23860
URL: https://issues.apache.org/jira/browse/CAMEL-23860
Project: Camel
Issue Type: Improvement
Components: camel-langchain4j-agent, camel-langchain4j-chat,
camel-langchain4j-tools
Reporter: Claus Ibsen
The camel-langchain4j-embeddings component already exposes token usage data
(input/output/total token count) and finish reason as exchange headers from the
LangChain4j Response object. However, the chat, agent, and tools producers
discard this metadata.
Specifically:
- camel-langchain4j-chat: extracts only response.text() and discards the
Response wrapper containing tokenUsage() and finishReason()
- camel-langchain4j-agent: same, no token data exposed
- camel-langchain4j-tools: reads finishReason() internally for control flow but
does not expose it as a header
The LangChain4j API provides this data in the Response object - the embeddings
producer already demonstrates the pattern:
{code:java}
if (result.tokenUsage() != null) {
message.setHeader(...INPUT_TOKEN_COUNT,
result.tokenUsage().inputTokenCount());
message.setHeader(...OUTPUT_TOKEN_COUNT,
result.tokenUsage().outputTokenCount());
message.setHeader(...TOTAL_TOKEN_COUNT,
result.tokenUsage().totalTokenCount());
}
{code}
The same pattern should be applied to the chat, agent, and tools producers so
users can track LLM token consumption, costs, and completion status.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
