fix(ai): redact local file paths from AI payloads#785
Merged
Conversation
Homeboy Results —
|
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Summary
file_info.file_pathfrom AI-visibledata_packetsJSON inAIStepWhy
Fetch handlers can attach
file_infowith local/internal file paths so downstream runtime steps can access files. That is valid internal pipeline state.The problem is that
AIStepwas serializing the fulldata_packetsarray directly into the model prompt. That exposed local paths likedm-files, temp paths, or legacy server paths to the model, which can then be copied into generated post content.This is not Reddit-specific. Reddit made the bug obvious on wire posts, but the exposure path is generic:
FetchHandler->DataPacket(file_info)->AIStepJSON serializationSo the safest fix is at the AI serialization boundary in core.
Testing
homeboy test data-machine