Repository navigation
chore: promote main to staging - #183
Merged
Merged
Conversation
- Added support for assigning sources to folders, enhancing organization and management capabilities. - Introduced new `folderId` property in various source-related files and updated relevant tests to ensure proper functionality. - Modified source route service to handle folder assignments, including validation and error handling for folder-related requests. - Updated tests across multiple domains to cover new folder assignment scenarios, ensuring comprehensive coverage. This enhancement improves the user experience by allowing better categorization of sources within the application.
- Added support for retaining evidence picks after retrieval, introducing a new `retainPicks` method in the ledger. - Implemented pending retention range management to track newly added evidence chunks. - Updated the agent harness runtime to prepare answer context based on retained evidence, improving response accuracy. - Refactored tests to cover new retention functionality and ensure proper integration with existing workflows. This enhancement improves the agent's ability to manage and utilize evidence effectively, leading to more accurate and contextually relevant responses.
- Introduced a new `resolveRetainedConnectedAssets` method in the ledger to handle connected assets for retained evidence, improving the assembly of answer context. - Refactored the `composeAnswerContext` function to support multiple asset types, including tables and images, ensuring accurate representation in responses. - Updated tests to validate the new asset resolution functionality and its integration with the answer context generation process. This enhancement improves the agent's ability to utilize connected assets effectively, leading to richer and more informative responses.
…ndling - Enhanced the `composeAnswerContext` function to include handling for cases where no knowledge base or memory results are found, providing clearer feedback in the generated messages. - Updated the test suite to cover new scenarios, ensuring that the context composition accurately reflects the absence of results and maintains the integrity of user interactions. - Refactored memory item formatting to improve clarity and consistency in the output. These changes enhance the agent's ability to communicate effectively when searches yield no results, leading to a more informative user experience.
- Removed the `referencedChunkCount` from the response structure to streamline data handling. - Updated tests to ensure that responses do not contain unwanted fields, specifically checking for provenance-only IDs and other irrelevant data. - Enhanced the logic for processing referenced chunks to focus on those with evidence, improving the accuracy of the citation pipeline. These changes improve the clarity and relevance of the data returned by the agent, leading to more precise responses in user interactions.
feat: refine notebook retrieval context and folders
Cite memories by mem:N in the prompt and resolve item ids in code so the model does not copy UUIDs. Co-authored-by: Cursor <cursoragent@cursor.com>
…handoff Connect Notebook chat to Memento session memory
Online chat can call Memento before it is deployed; keep the turn going and mark those misses for later. Co-authored-by: Cursor <cursoragent@cursor.com>
Treat unreachable memory search and send as warnings
Persist the Notebook agent TRACE on chat messages and include it with the last five turns so Memento can extract error and reflection after a complaint. Also limit Knowhere search to once per turn. Co-authored-by: Cursor <cursoragent@cursor.com>
…trace Send current-turn TRACE and recent turns with memory capture
Place figures at the chunk placeholder, skip unreachable image/table URLs with a warning, and keep the default chat model on DeepSeek V4.1 Flash. Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
…mages 作答时把图内联成 base64，不再把 S3 链接交给模型
Stop composing tables and images locally. Answer context now eats the whole Knowhere evidence package. Co-authored-by: Cursor <cursoragent@cursor.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
…e-consume feat(chat): consume Knowhere assembled evidence for answers
Co-authored-by: Cursor <cursoragent@cursor.com>
…inject feat(chat): inject profile, lessons, and crystal ranking factors
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
This branch was successfully deployed
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
mainontostaging, including profile/lessons inject and earlier evidence consume work.Test plan
Made with Cursor