Repository navigation
Redesign the research workspace and integrate source-bound AI analysis - #303
Merged
Merged
Conversation
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
1 participant
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The research graph, execution output and manuscript workflows now share a focused project workspace. Five primary destinations and contextual graph/run actions preserve the existing scientific capabilities; the graph stays visible beside the shared Paper and Analysis panels.
The new right-side AI analysis integrates project, selected-node and path scopes with the actual API, scheduler, worker and ModelClient. It gathers recorded graph/configuration/dependencies, experiment history, metrics/errors/artifacts, linked research resources, tool/event/decision history and bounded material text. Each explicit request keeps its reviewed revision/fingerprint and UUID. Saved reports expose experiments, problems, findings, recommendations, server-owned literature references and source coverage, with historical/stale-source and changed-artifact handling. Node selection only changes focus; advice does not alter experimental state.
Draft guards retain later edits across refresh, navigation, conflicts and delayed acknowledgments. The change also repairs full-ID run/path navigation, responsive graph actions, runtime history preservation, file freshness, source-bound figure execution and PDF preview cancellation.
Validation on the delivered production assets:
Live model scientific advice and full-text citation support have not been evaluated. Linux Agent/human-decision behavior, dedicated PostgreSQL, complete accessibility and performance benchmarks retain their documented limits. The original macOS backend baseline was 1429 PASS / 71 SKIP / 63 FAIL; failure classifications and the subsequently repaired skill-catalog assertion are recorded. Publication-profile updates still rely on a fresh-read check rather than an atomic server revision guard.
Full delivery report · Per-case results and evidence · Acceptance matrix