Export Research Bundle
Purpose
Auto-assemble a read-only Research Bundle in Dashboard Products from existing Zotero library and Synthesis context, based on a declared paper intent. The bundle collects relevant topics, core papers, and related papers with their available analysis artifacts.
Inputs
| Parameter | Required | Description |
|---|---|---|
paperTitle | Yes | Working manuscript title used to find research materials. |
researchContent | Yes | Research problem, methods, scope, and intended contribution. |
articleType | No | Manuscript type (default: original research). |
maxTopics | No | Maximum number of relevant topics to include, range 0–5 (default: 5). |
maxCorePapers | No | Maximum number of core papers, range 1–20 (default: 20). |
maxRelatedPapers | No | Maximum total related papers including core, range 1–80 (default: 80). |
No Zotero item selection is required.
Behavior
- Receive the paper intent parameters from the user.
- Discover candidate materials from existing Synthesis Topics, Zotero library items, and available citation graph context.
- Perform bounded evaluation to distinguish core papers from related papers.
- Assemble the Research Bundle with topic reports, bibliographic metadata, and available v2 analysis artifacts (digests, references, citation analyses, conversation content).
- For core papers, prefer Markdown source with local images; fall back to PDF; record a warning if neither is available.
- Register the bundle as a read-only product in Dashboard Products.
Topic, graph, analysis artifact, or source unavailability degrades gracefully — the workflow continues with whatever evidence is still readable and records diagnostics and warnings. If no papers meet the criteria, the run ends without registering a product.
Output And Apply
The Research Bundle is registered in Dashboard Products as a read-only artifact. Its structure:
| Path | Description |
|---|---|
README.md | Agent- and human-facing entry point with suggested reading order, file naming, topic/paper index |
manifest.json | Machine-readable inventory of v2 artifact paths, provenance, file integrity, and diagnostics |
topics/<topic-id>/report.md | Topic synthesis report (when available) |
papers/<paper-id>/metadata.json | Portable bibliographic metadata per paper |
papers/<paper-id>/source.md | Markdown source (when available) |
papers/<paper-id>/digest-*.md | Literature Analysis digest artifacts (when available) |
Only topics/ and papers/ semantic directories are used alongside the root files. Markdown images are included only when their resolved local path falls within the Markdown file's directory tree; out-of-tree or missing images retain their original links but are not registered as product files.
Estimated Duration
Depends on library size, candidate count, topic/graph availability, and backend response speed. Progress and results are visible in the run panel.
Model Recommendation
A model with strong semantic understanding and tool-calling ability is recommended. The task requires judging topic and paper relevance against the paper intent and correctly using read-only Zotero and Synthesis context.
Dependencies
- Backend: Skill-Runner
- Skill:
export-research-bundle - Host Bridge: Requires permission to read Zotero and Synthesis context
Related Workflows
- Literature Analysis — Generate digest and citation analysis artifacts that can be included in the bundle
- Literature Search & Ingest — Search and ingest missing literature before assembling the bundle
- Export/Import Literature Bundle — Export portable ZIP bundles of Zotero items (different purpose)