Artifact Inspection Guide

On this page

Inspect sources, licenses, scans and hosted files. These records do not confirm that a model will run.

First: make one live call

This public catalog search needs no key. Inspect a returned artifact before recommending or downloading a file; a successful search can be empty. Read the current download-plan before deciding whether a file is downloadable. Keep path, downloadUrl and sha256 from the same ready file row.

curl -fsS 'https://huggingbay.xyz/api/mcp' -H 'Content-Type: application/json' -d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"search_artifacts","arguments":{"q":"qwen","limit":5}}}'

REST contract for agents: /api/openapi.json (OpenAPI 3.1). Copy-paste setup for curl, Python, MCP and Claude or ChatGPT tools: agent quickstart.

Slash commands

Name the MCP server hugging-bay and start with /mcp__hugging-bay__find-model-for-my-machine. Use prompts/list to pick a command for hardware, download checks, model comparison, licenses or requests. Fill the arguments with completions; request commands ask for your OK before submitting.

Run instructions are unavailable here. Check the exact file and run evidence.
Self-reported reviews, receipts, votes, reports, and benchmark evidence have no public decision effect.

Markdown version JSON version

Quick answer: compatibility and sources

How should an agent inspect a model record before recommending a file?

Read the model card, exact file list, license and availability status, then cite a recorded checksum and any accepted run receipt. Recommend only the file supported by those records. A card or matching checksum gives limited evidence; neither establishes safety, compatibility, speed or answer quality.

A model card summarizes a model. A file record describes one downloadable file; a run receipt records the outcome for specific file bytes and a named setup.

The OpenAPI schema defines the card, file and evidence endpoints. Agent discovery metadata lists available interfaces. As a concrete example, open the Qwen3-8B record, its exact file list, and its run evidence. This cookbook does not create a run receipt.

Updated 2026-10-21.

Page sources

Page-copy source revision 8871bab84e1cf80aea1686a944ec335078a6c036 · Source files: apps/web/server.mjs, apps/web/lib/aeo-page.mjs, apps/web/lib/api-status-contract.mjs, apps/web/lib/agent-readable.mjs, apps/web/lib/docs-hub-model.mjs, apps/web/lib/compare-pages.mjs · Full revision record

Key facts

Catalog vs runtime: Metadata access, file hosting and a tested load are separate checks.
Checksum: SHA-256 confirms file identity, not license permission or answer quality.
Receipts: Run evidence is scoped to one file and setup; it does not transfer to other files.

Evidence you can inspect (plain text)

GET /openapi.json lists request and response shapes for public API reads.
GET /api/agent-discovery lists agent-facing interfaces when enabled.
Open /artifact/hf-model-qwen-qwen3-8b/files for exact filenames, availability and any recorded SHA-256 for the model.
Open /evidence/artifacts/hf-model-qwen-qwen3-8b for run-readiness evidence tied to that record; absent receipts stay unmeasured.
Open /api/artifacts/hf-model-qwen-qwen3-8b/card.md for machine-readable source facts (not a runtime test).
These documents describe interfaces and catalog reads; they do not execute or score models.

Unmeasured states are stated explicitly; this page does not invent run scores or compatibility guarantees.

Which date applies to which fact

Committed page-copy sources: 2026-10-21
Upstream model metadata: Dates on artifact records when shown; not implied by page-copy revision
Run receipts: Only on /evidence pages when recorded; otherwise unmeasured

Page-copy revision (Updated line above) identifies committed text sources, not upstream model edits or test outcomes.