Skip to content

Job Stories MCP-Server-Portaljs next steps #1456

@Abeelha

Description

@Abeelha

UPDATE 10/08/2025 - Created PR with all Tools added:

  • [ADD]: get_dataset_stats, preview_resource, get_related_datasets, get_organization_details, compare_datasets;

PR here: datopian/portaljs-mcp-server#3


Job Stories - PortalJS_MCP_Server - User Perspective (Non-Technical Users)

Regarding our current implementation of mcp-server-portaljs, in the following repo: https://github.com/datopian/portaljs-mcp-server

Context

  • Current implementation has only Fetch and Search functionality tools for Claude-desktop / ChatGPT Connectors, as shown in the image bellow:
Image
  • These job stories represent needs from students, researchers, and general users who want to work with data for analysis, research, and exploration.

  • What can we implement to make the MCP-Server better? Bellow we will be having job stories and their value for future development.


Job Story 1: Quick Dataset Overview for Research

  • When I'm starting a research project and need to quickly understand what data is available on a specific topic,

  • I want to see a summary of multiple datasets at once including their size, freshness, and key themes,

  • So that I can decide which datasets are worth exploring further without having to open each one individually.

Context

Students and researchers often need to scan through dozens of potential datasets. Opening each one to understand its contents wastes time when most won't be relevant.


Job Story 2: Understanding Data Before Committing

  • When I find a dataset that might be useful but I'm unsure if it contains the specific information I need,

  • I want to preview the structure and sample values of the data files,

  • So that I can confirm it has the right columns and data types before investing time in downloading and analyzing it.

Context

Non-technical users often struggle to understand if a dataset will meet their needs just from descriptions. They need to "peek inside" without technical tools.


Job Story 3: Finding Related Information

  • When I'm working with a dataset but realize I need additional context or complementary data,

  • I want to discover other datasets from the same source or related to similar topics,

  • So that I can build a more complete picture for my analysis or presentation.

Context

Research and analysis often require multiple data sources. Users need to explore connections between datasets organically.


Job Story 4: Verifying Data Reliability

  • When I'm preparing to cite data in academic work or need to justify my data source to stakeholders,

  • I want to quickly access information about who published the data, when it was last updated, and what organization maintains it,

  • So that I can assess its credibility and provide proper attribution in my work.

Context

Academic integrity and data trustworthiness matter. Users need confidence in their sources without digging through technical metadata.


Job Story 5: Comparing Dataset Options

  • When I've found several datasets on the same topic and need to choose the best one for my specific use case,

  • I want to compare their coverage periods, update frequency, and scope side-by-side,

  • So that I can make an informed decision about which dataset best fits my timeline and requirements.

Context

Multiple datasets on similar topics exist. Users waste time downloading and comparing when they could filter earlier in the process.


Implementation Considerations

These stories have the potential enhancements:

  • Dataset summary/statistics endpoints
  • Resource preview capabilities
  • Related dataset discovery
  • Organization/source filtering
  • Comparative metadata views
  • Freshness/update indicators

All achievable with GET operations on existing data structures, no need for having POST/PUT, and other CRUD actions since the MCP is for open-to-use and authless.

Metadata

Metadata

Labels

No labels
No labels

Type

No type

Projects

Status

Done

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions