huggingface / huggingface/hf-discover

Federation request: index the iParq Parquet Inspector catalog

Open
#44 0 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
40
Forks
10
PR merge metrics
No merged PRs in 30d

Description

## Request

Please consider indexing or federating the canonical iParq ARD catalog in Hugging Face Discover, or advise which publisher-ingestion path Hugging Face prefers.

iParq is an open-source, local, read-only CLI and Agent Skill for inspecting Parquet storage metadata. A practical Hugging Face use case is inspecting downloaded Dataset Hub Parquet shards without uploading their row data elsewhere.

## Live discovery surfaces

- ARD catalog: https://iparq.dev/.well-known/ai-catalog.json
- Identifier: `urn:air:iparq.dev:skill:parquet-inspector`
- Agent Skill: https://raw.githubusercontent.com/MiguelElGallo/iparq/main/.agents/skills/iparq-parquet-inspector/SKILL.md
- Documentation: https://iparq.dev/docs/
- Source: https://github.com/MiguelElGallo/iparq
- PyPI: https://pypi.org/project/iparq/

The catalog serves ARD `specVersion` 1.0 over HTTPS with `Content-Type: application/json` and permissive CORS. The current catalog entry is version 0.7.0 and includes capabilities, tags, and five representative queries.

## Verified direct navigation

The current `hf-discover` client resolves the catalog and returns the iParq Skill:

```bash
uvx --from hf-discover hf-discover navigate \
https://iparq.dev \
"inspect parquet compression and bloom filters" \
--json
```

However, the hosted Hugging Face Discover search currently returns no iParq result for an exact `iparq` query.

## Questions

1. Can the hosted Hugging Face Discover registry ingest or federate a static external publisher catalog such as this one?
2. If arbitrary publisher ingestion is not supported yet, what is the preferred registration path?
3. Is contributing the Skill to `huggingface/skills` the intended interim route, or should that repository remain limited to core Hugging Face functionality?

A separate curated-Skills contribution is open at https://github.com/huggingface/skills/pull/206. This issue is specifically about external ARD catalog discovery/federation rather than inclusion in the Hugging Face Skills bundle.

Happy to adjust the catalog or provide a conformance fixture if useful.

Contributor guide

No contributing guide indexed for this repository

Research direction

Start by running the documented hf-discover navigate command and comparing it with the hosted exact `iparq` search result. Then inspect the client/server entry points that handle catalog discovery and registry search; the issue is complete when the supported federation or registration path is established, with any needed conformance fixture identified.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
api, developer-experience
Issue type
Feature
Difficulty
5/5
Estimated time
Over a week
Activity status
Quiet
Clarity
Needs clarification
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.