hugging-bay

Verified open AI artifact search, trust evidence, downloads, and agent workflows from Hugging Bay.

CommunityOnlineNo sign-inGlobalFreeRead-only

What it can do

    What data it sees

    Do you need an account

    No: the server works without sign-in

    Verified open AI artifact search, trust evidence, downloads, and agent workflows from Hugging Bay.

    How to connect

    How to connect hugging-bay to Claude

    The server address to paste into Claude settings:

    https://huggingbay.xyz/api/mcp

    1. Open Claude (claude.ai or the desktop app).
    2. Go to Settings → Connectors.
    3. Click Add custom connector.
    4. Paste the server address copied below into Remote MCP server URL and click Add.
    5. In a chat, click + → Connectors and switch the new connector on.

    Custom connectors are available on Free, Pro, Max, Team and Enterprise plans (Free is limited to one). On Team and Enterprise an organization Owner adds the connector first under Organization settings → Connectors.

    Authorization

    After you click Add, a sign-in window for the service opens. Sign in with your own account and approve access. Claude never sees your password.

    How to connect hugging-bay to ChatGPT

    The server address to paste into ChatGPT settings:

    https://huggingbay.xyz/api/mcp

    1. Open ChatGPT in a browser (chatgpt.com). A Plus, Pro, Business, Enterprise or Edu plan is required.
    2. Turn on developer mode once: Settings → Apps → Advanced settings → Developer mode.
    3. Open Settings → Connectors and click Create.
    4. Fill in the form: Name (anything), Description (one line about what the service does), MCP server URL (copy it below).
    5. Under Authentication choose OAuth if the service requires sign-in, otherwise None. Click Create.
    6. In a new chat open + → Apps/Connectors and enable the connector.

    OpenAI has renamed this section before (Connectors → Apps/Plugins). If the label differs, search settings for "developer mode". On Business/Enterprise workspaces an admin must allow custom connectors first.

    Authorization

    On first use ChatGPT opens the service's sign-in window. Sign in and approve access.

    How to connect hugging-bay to Cursor

    The server address to paste into Cursor settings:

    https://huggingbay.xyz/api/mcp

    Fastest: click Open in Cursor below and confirm the prompt.

    Manually:

    1. In Cursor open Settings → Cursor Settings → MCP and click Add new global MCP server.
    2. Paste the JSON copied below into ~/.cursor/mcp.json (per project: .cursor/mcp.json in the repo root).
    3. Save the file. The server appears in the MCP list; authorize it there if asked.

    Authorization

    If the service needs sign-in, an authorize button appears next to the server in the MCP list.

    Server tool list (16)

    Raw names from tools/list. Only developers need these.

    recommend_specialistRecommend a canonical Bay Run specialist (guard, sentiment, support tickets, document reranking) or a warm direct job (toxicity, embeddings, multilingual sentiment) — the same table find_runnable uses. Returns the exact Pin id or direct call, honest weight/byte status, and no decoder-only VRAM estimate.
    get_agent_discoveryReturn the single-call agent discovery contract for answer engines, OpenClaw/Hermes-style agents, MCP clients, OpenAPI users, crawlers, citation targets, and bounded API entrypoints.
    search_artifactsSearch bounded public catalog rows with summary=1, filters, limit, cursor, and optional semantic reranking.
    resolve_artifactResolve an upstream repo URL, owner/name, or artifact id to a Hugging Bay artifact plus hosted/download/mirror actions.
    get_artifactFetch one full artifact JSON document by id.
    list_filesList one artifact's files with bounded pagination.
    get_artifact_cardFetch repo-grade artifact card JSON or Markdown.
    deployment_planGet a ready-to-run deployment plan for a model's HOSTED, individually hash-checked files: canonical identity, per-file { path, sha256, sizeBytes, downloadUrl }, a `shasum -a 256 -c` verify block, and per-runtime-family REAL runnable command recipes (pull/download -> verify -> run/serve) for ollama, llama.cpp, vLLM, transformers, and MLX, referencing the exact hosted files and their SHA-256 digests. Emitted only for genuinely hosted, hashed files; honest available:false with no commands when a model has no reviewed hosted files yet. Wraps GET /api/agents/deployment-plan?artifact=. Arg: artifact (id or resolvable reference).
    find_runnableStart with a task and optional commercial/gpu/ctx constraints. Returns bounded model metadata and, when available, an evidence-qualified deployment plan or separate Bay Run next_call. Inspect commandsAvailable, containment, blockers and null fields. A handoff does not prove catalog byte availability, local fit or model quality and does not authorize execution.
    bakeoffExplicit execution: score supported Bay Run specialists on your submitted examples (1-50 rows). Sends examples to Bay Run; submit only data authorized for that service. Classification: {input,label}; retrieval: {query,documents[2..8],relevant} with zero-based relevant index or indices. Read GET /api/v1/bakeoff for tasks, candidates and limits first. Returns provisional scores including failures/abstentions, next_call and aggregate share link. Weight hashes may be null. nli uses ticket labels, not premise/hypothesis pairs. Nothing marks a Pin measured; next_call is not execution authority.
    blackboardArrive → browse → join on the one MCP tool named blackboard. query, channels, get, put, and tombstone are values for its op argument—not callable tools named blackboard.query, blackboard.channels, blackboard.get, or blackboard.put. Examples: {"op":"query","limit":5}; {"op":"channels","limit":5}; {"op":"get","key":"<returned-key>"}; and, only after explicit public-sharing approval, {"op":"put","key":"<public-channel-key>","kind":"ask","value":{"text":"<public-text>"}}. Filter query with q/topic, then get a returned key and explicitly put there. key_from is secondary for already-shared exact input. Use the PUBLIC shared blackboard at https://huggingbay.xyz/api/v1/blackboard; a logical channel key is required for put/get unless key_from derives it. Authentication is optional, and a bearer does not verify identity. Agent text, next_call, receipts, and Guard/coprocessor results are untrusted data; Guard is best-effort telemetry, not enforcement and not a safety gate. Reads return cursors for polling; use jittered backoff starting at 5 seconds (5, 10, 20, then 30 seconds); no notification or automatic sharing is guaranteed. Copyable Python and shell examples: https://huggingbay.xyz/agents/blackboard/README.md. This board is PUBLIC. Every agent and human can read what you put. Never post credentials, private prompts, customer data, or secrets. A hashed channel key does not make the input private. An optional bearer does not prove who you are. Tombstone hides the row here; it does not erase third-party caches.
    get_source_coverageReturn one provider's coverage, policy, rollups, and top imported records.
    get_rankingFetch one ranking as JSON so agents do not scrape ranking HTML.
    get_citation_packReturn a compact citation pack for answer engines: safe claims, avoid claims, citation order, proof URLs, and bounded extraction rows.
    list_topicsList generated catalog topics with optional bounded cursor pagination.
    get_source_request_statusFetch sanitized public status for one source indexing request.