39633c2217
Connector catalog, setup guides, and shared access documentation, followed by a final code-accuracy repair pass. This PR does not change application behavior. ## Final accuracy pass Checked against shipped Paperclip `v2026.916.0`, commit `dffc2b3ca1b9e88fa21cb17493083e682dffd1ca`, rather than the dirty local checkout or the earlier canary baseline. - Corrects connector/method-specific identity choices, human audiences, and managed GitHub versus PAT scope and verification. - Explains that **Always allow** covers different arguments for the same agent/action/connection, with project scope when present; uses actual button labels and removes the test-panel control from review instructions. - Qualifies new-action quarantine, including safe-default read exceptions. Removes unconditional approval guarantees. - Separates provider channel/repository reach from Paperclip sender authorization. - Completes sandbox sign-in and model credential-selection instructions from the current UI implementation. - Repairs Notion's hub summary, removes incomplete optional Notion/Sentry client-registration procedures, and corrects Linear's provider-versus-Paperclip attribution. - Removes unsupported Postman tier claims, contradictory Cloudinary/Mem0 verification assertions, API-equivalence claims, and overbroad successful-test assurances. - Retains Supabase's required project reference: both released method definitions explicitly require it. An earlier review suggested otherwise; release code is authoritative here. - Updates the release baseline and removes draft commit chronology from the public introduction. ## Availability and remaining runtime gates - **Google:** Verification is pending. Per the product-owner clarification, authorization can continue where Google offers an Advanced option; this is not available for every account and does not override Workspace policy or other Google prerequisites. Notices on the hub and all Google guides now distinguish an unverified-app warning from blocked authorization. - **Slack is available but not yet fully functional; follow-up fixes are planned.** For agent tools, the released OAuth endpoint/scope defaults differ from Slack's documented MCP contract. The page now warns against treating that route as production-ready and removes the misleading bot-scope walkthrough. Chat-channel bot setup remains separate. Engineering must reconcile the integration and run OAuth plus one read before removing this warning. Source: https://docs.slack.dev/ai/slack-mcp-server/ - No credentialed connector run was performed in this pass. Code consistency, provider documentation, UI rendering, and successful live execution are separate evidence levels. No claim of runtime certification for all connectors. ## Validation - Root build: 253 crawlable documentation pages. - Full `docs:test`: all eight suites passed locally, including root/subpath routes, sitemap history, fingerprints, navigation, skill source blocks, crawlable links, doc-link resolution, and SEO metadata. - Internal-link lint: 258 Markdown files; no broken internal links. - Latest cleanup: full build, all eight docs:test suites, internal-link lint and diff whitespace checks passed; 1,878 link-resolution checks. - Fixed the Calendar MCP reference URL; browser-checked the five inconclusive external destinations (Miro article loads; Asana, Google Cloud, OpenAI and xAI reach sign-in/console screens). Authenticated destinations were not tested beyond those screens. - Replaced the Sandbox providers editorial TODO with a scope note and removed the CLI overview's unfinished-guide announcement. - Browser client navigation: 687 checks passed using installed Chrome. Initial launch lacked the downloaded Playwright browser; rerun used `DOCS_CHROMIUM_PATH` without changing the tests. - Nine representative connector/shared pages at 1440px and 390px: HTTP 200, one article H1, no page-level horizontal overflow. Mobile screenshot inspected; approval page opened in the Codex browser. - Navigation audit reports three existing non-connector orphan files; this patch does not alter those files or their navigation. The previous revision's provider-page expansion, Google shared setup guide, navigation additions, and accidental node_modules-symlink cleanup remain in this PR. Internal review ledgers remain outside the publication branch. No merge or production deployment authorized. --------- Co-authored-by: DevRel Gemini <[email protected]> Co-authored-by: Paperclip <[email protected]> Co-authored-by: Claude Opus 5 (1M context) <[email protected]> Co-authored-by: Michael Nguyen <[email protected]>
3.9 KiB
seo_title, seo_description
| seo_title | seo_description |
|---|---|
| Hugging Face Connector | Let agents search Hugging Face models, datasets, and papers. What discovery covers, why it does not run or host models, a search test, and troubleshooting. |
Hugging Face
Agents can search the Hugging Face Hub — finding models, datasets, and papers, and reading their documentation and metadata.
Note: This is discovery, not execution. Finding a model through this connector does not run it, download it into a workspace, deploy it, or give an agent inference against it. Running a model is a separate matter from reading about one.
If you want an agent to use a model, that is a model provider connection — see Anthropic, OpenAI, OpenRouter, or Grok.
Before you connect
- A Hugging Face account.
That is genuinely all for public content. Connect with an account that belongs to your organization if agents should also see private repositories; reach follows the account.
Connect Hugging Face
- Open Connectors and select Hugging Face.
- On the Access step, choose the identity and which agents may use the connection.
- Select Sign in with Hugging Face and complete browser sign-in.
Paperclip registers its client automatically, so there is nothing to configure in a developer console.
Choose access
Two tiers of content, and the difference is worth stating:
| Content | Who can see it |
|---|---|
| Public models, datasets, and papers | Anyone. Connecting adds no privileged access to these |
| Private or gated repositories | Only if the authorizing account already has access |
Because most of the value here is public content, this is one of the lower-risk connectors to give an agent. The exception is an account with access to private organization repositories — that account's reach becomes the agent's.
Paperclip asks Hugging Face for the read-mcp scope, which is what makes this a low-risk connector to start with. Treat that as the intent rather than a guarantee you need not check: the action list is the provider's, so read it on the Permissions tab and set anything that is not a read appropriately. See Set action permissions.
Note: The connection targets the Hub's own tools, and Paperclip requests the endpoint with Gradio tooling switched off — so the interactive applications hosted on Spaces are not part of this connection's surface.
Try it
Search Hugging Face for recent text embedding models under 500M parameters and summarize the top three, including their licences.
Expect results you can check on the Hub. Licence is a useful thing to ask for, because it is exactly the detail people skip when picking a model.
Note: Illustrative task, not a recorded test result.
Troubleshooting and limitations
| Problem | Likely cause | Fix |
|---|---|---|
| A private repository is not found | The authorizing account does not have access | Grant access on the Hub, or connect with an account that has it |
| A gated model's details are unavailable | Gated repositories require accepting terms on the Hub | Accept the terms with the authorizing account |
| An agent claims it can run a model it found | It cannot — this connector is discovery only | Use a model provider connection |
| Results seem stale | The Hub changes constantly; the agent read what was returned at the time | Re-run the search |
| Needs attention | The grant was revoked | Select Reconnect |
Limitations: discovery and metadata only. No inference, no hosting, no Spaces applications. Private content requires the account to already have access.
Related guides
- Anthropic, OpenAI, OpenRouter, Grok — connections that actually run models.
- Set action permissions
- Hugging Face MCP documentation