feat(rag): expose reasoning trace via public run_with_trace()#39
Merged
ceberam merged 1 commit intoJun 25, 2026
Merged
Conversation
Contributor
|
✅ DCO Check Passed Thanks @pjmalandrino, all your commits are properly signed off. 🎉 |
Contributor
Merge Protections🟢 All 2 merge protections satisfied — ready to merge. Show 2 satisfied protections🟢 Enforce conventional commitMake sure that we follow https://www.conventionalcommits.org/en/v1.0.0/
🟢 Require two reviewer for test updatesWhen test data is updated, we require two reviewers
|
Open
12 tasks
Codecov Report✅ All modified and coverable lines are covered by tests. 📢 Thoughts on this report? Let us know! |
Add RAGTrace model (query, per_document, final_answer) and a public run_with_trace() method on DoclingRAGAgent returning the full per-iteration trace that was previously discarded inside run(). run() becomes a thin wrapper around run_with_trace(): single source of truth for the loop, no behavior change for existing callers. RAGTrace, RAGResult and RAGIteration are re-exported at package level so downstream consumers can type-hint against the trace. Closes docling-project#26 Signed-off-by: Pier-Jean Malandrino <pierjean.malandrino@scub.net>
pjmalandrino
force-pushed
the
dev/rag-run-with-trace
branch
from
32622b4 to
f72f877
Compare
Contributor
Author
|
Hey @PeterStaar-IBM / @ceberam , any update on this ? This is necessary for my visualization layer on Docling-Studio. |
ceberam
approved these changes
Jun 25, 2026
ceberam
left a comment
ceberam
left a comment
Member
There was a problem hiding this comment.
Thanks @pjmalandrino for suggesting this PR.
I'm happy with the design and implementation of this feature. Actually, we thought of adding a similar tracing capability across all agents that could be exported to a file for debugging purposes. However, I'm fine with this solution for the RAG case for now, specially if it helps with your integration.
pjmalandrino
added a commit
to pjmalandrino/docling-agent
that referenced
this pull request
Jun 26, 2026
…a file Generalize the run_with_trace pattern from docling-project#39 into a generic, typed AgentTrace that every agent produces, and let the orchestrator compose the traces of the sub-agents it dispatches to into a tree that can be exported to a single JSON file for debugging. - New AgentStep / AgentTrace value objects (agent_trace.py). AgentTrace carries ordered steps, nested children (sub-agent traces), timing, model id and the produced document on `output` (excluded from serialization; result_name is the persisted pointer). SerializeAsAny on children preserves subclass fields. - BaseDoclingAgent.run_with_trace(): concrete default so every agent exposes a trace (timing + model + result) with no bespoke code; agents override to add steps. run() stays the source of truth for the document. - RAGTrace becomes a subclass of AgentTrace, so docling-project#39 is preserved (same fields, same construction, covariant return) and a RAG run nests into the tree. DoclingRAGAgent builds the answer doc onto output; run() returns it. - DoclingOrchestratorAgent.run_task_with_trace() composes the tree by recording each dispatched sub-agent trace; run_task() is unchanged and incurs no overhead. - LoggingConfig.trace_path + CLI export. Public re-exports: AgentTrace, AgentStep. No global state, no logging coupling: the trace is a returned value object. Design doc: docs/design/37-export-session-trace.md. Closes docling-project#37 Signed-off-by: Pier-Jean Malandrino <pierjean.malandrino@scub.net>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Add RAGTrace model (query, per_document, final_answer) and a public run_with_trace() method on DoclingRAGAgent returning the full per-iteration trace that was previously discarded inside run().
run() becomes a thin wrapper around run_with_trace(): single source of truth for the loop, no behavior change for existing callers.
RAGTrace, RAGResult and RAGIteration are re-exported at package level so downstream consumers can type-hint against the trace.
Closes #26