d627632874
Docs for parent release `v2026.1106.0` (`d69631b0`), up from `v2026.1023.0` (`1b4c8038`): 515 parent commits. ## How this branch was built The tag is 914 commits behind what `nightly` has drafted, so this branch isn't cut from `nightly` HEAD. It starts at nightly commit `1952052` (parent `043f2514`), 19 commits before the tag, then adds: - the `main` hotfix (#160) - the sync tooling from `nightly` - docs for those 19 commits Later nightly drafts stay on `nightly`, so no pages had to be removed. No `nightly-draft/*` PRs are open. ## What changed - **Runner by default:** new Claude, Codex, OpenCode and Pi agents start on Paperclip Runner on supported hosts, with **Legacy runner** and **Upgrade runner**. - **Workspaces:** cleanup of finished workspaces, optional Git expiration, and restore points for unsaved work. Restore points are off by default. - **New pages:** Prompt injection prevention, Google Workspace, GitHub Copilot, Speko, Braintrust, Daytona, Ditto and Unikraft. - **Breaking and upgrade notes:** the run history limit, Sentry run failures off by default and the DSN variables, the Docker start command for custom images, `--yes` on `instructions-file:delete`, and slower migrations on large instances. - **Also covered:** project repositories as skill sources, roster bulk edit, Artifacts V1 (experimental, off by default), composer dismissal, `expectedStatusVersion`, `unexecuted_review_stage`, and plugin workspace providers. - **Changelog:** a v2026.1106.0 entry in `reference/changelog.md`. ## Older text corrected These were wrong at the tag, and most are also wrong on the live site: - which permissions let an agent read a peer's configuration or change instructions (`api/agents.md`) - `decideNowCount`, which doesn't exist; the field is `deskBadgeCount` - `file-resources` routes, which are board-only - the **Create project** dialog fields, a CEO rule the code doesn't enforce, and where a terminated agent's reports appear - `instructions-file:put` and `:delete` examples that always failed (`cli/agent.md`) - a curl example that returned `422` (`api/cases.md`) - Google Workspace **Reconnect** instructions, which claimed it reopens the service picker - the `permissionKey` table, which listed 8 of 21 keys - drafting text at the end of `api/agents.md` ## ⚠ Please check these - **Product behaviour that may be a parent bug.** The docs now describe the tag's code, which you may prefer to fix upstream: - The project **Environment** dropdown saves a choice, but runs ignore it (`experimental/environments.md`). - `environments:manage` is granted by roles but no environment route checks it. - `suggest-changes` is documented as agent-only, because only agents have a consent path. Company skill policy decides skill edits. - **Reader-visible mismatches in the parent UI.** The docs follow the code. Upstream copy still mentions a "Built-in agents tab", says Copilot "awaits qualification", and tells people to "Reconnect Google Workspace to change access". ## ⚠ Post-tag leaks removed Nothing to remove from an already-drafted page: post-tag surfaces such as 1Password, Vercel, Supabase deploy, custom gateway onboarding and Artifacts V1 as default never came onto this branch. Authors left out post-tag hunks of the next nightly draft, and the reasons are in the run notes. ## ⚠ Drift None. Zero findings at the tag, from a full parent snapshot. ## ⚠ Verification Failures The tag-accuracy gate reports one leak and nine review entries. None is a real failure: - **Leak:** `reference/cli/service.md`, `paperclipai payload and`. This is a false positive; it's prose from a real message at `cli/src/commands/service.ts:184`. - **Review, 8 new connector pages:** the gate can't confirm new pages. Reviewers confirmed each surface exists at the tag. - **Review, `local_tools`:** an example value from the tag's Runner README. Not provable from code, and kept deliberately: - `sandbox-providers.md` uses `sandbox.yaml`. The tag's README says `install.yaml`, which returned 404 when checked. The upstream fix is post-tag. - The Google Workspace row about older OAuth brokers rests on release notes and a design doc. The broker isn't in this repo. - External claims about provider behaviour (for example Braintrust, Daytona, Unikraft and Google scopes). ## ↻ Renames detected None. The rename helper wasn't run: the 19-commit delta adds no new adapter, skill, or plugin directories that it would check. ## Verification All checks read a checkout pinned to the tag. - **First round:** authors, a release-notes audit, an audit of corrections stranded on `nightly`, an audit of 321 lines the last realign dropped from `main` (about 130 checked directly in code), and six fresh verifiers. - **Second round:** seven new verifiers re-checked every changed area, tracing behaviour rather than matching labels, and fixed about 50 more defects. That volume means more may remain. - **Limits of that second round:** - Each verifier reported what it only string-matched. That includes parts of Artifacts V1, the **Upgrade runner** notice, Copilot run behaviour, and some Plugin SDK contract text. - The second-round fixes themselves weren't reviewed by a third round. - **Gates:** `docs:build`, `docs:test` and `sync:check` pass. - **Nothing was run against a live instance.** All of this comes from reading code. ## Screenshots **196 of 278 screenshots are stale and none were recaptured.** The list is in `SCREENSHOTS_PENDING.md`. Some visibly differ from the text, such as the removed **Run with provider trace** button and the old roster tabs. ## After merge 1. Keep the PR title as is: `tag-docs-release.yml` reads the version from it. 2. Confirm the `docs/v2026.1106.0` tag exists. 3. Run `node scripts/sync/realign-nightly.mjs release/v2026.1106.0 --push`, then audit for dropped lines and duplicated paragraphs, and set `nightly` state back to `branch_mode: nightly`. 4. Check that `docs.paperclip.ing` serves the release: a new page, a new claim, and the changelog's top entry. 🤖 Generated with [Claude Code](https://claude.com/claude-code) --------- Co-authored-by: Claude Opus 4.8 (1M context) <[email protected]> Co-authored-by: DevRel Gemini <[email protected]> Co-authored-by: Paperclip <[email protected]> Co-authored-by: Dotta <[email protected]> Co-authored-by: ClaudeCoder <[email protected]>
34 KiB
paperclip_version, seo_title, seo_description
| paperclip_version | seo_title | seo_description |
|---|---|---|
| v2026.1106.0 | Paperclip Connectors | Every service Paperclip can connect, grouped by what it is for, with the access model, action permissions, and a setup guide behind each one. |
Connectors
A connector is a saved connection to an outside service: a Gmail mailbox, a GitHub organization, a Notion workspace, a PostHog project. You set one up once, decide who and what it is for, and Paperclip carries that decision into every run.
Note: These guides cover the connector model in Paperclip v2026.1106.0. Available methods depend on your installed version, instance configuration, and provider approval. A catalog listing does not by itself establish that provider sign-in is available.
Warning: Google verification pending. Paperclip has not yet completed Google app verification. You may see an unverified-app warning during authorization. If Google offers an Advanced option to continue to Paperclip, you can choose to proceed after reviewing the requested access. This option is not available for every account; Workspace administrator restrictions and other Google access requirements still apply. Contact [email protected] if you cannot connect.
Note: New Google accounts connect through the single Google Workspace connector, where you choose Gmail, Drive, and Calendar access in one setup. The nine single-service Google connectors are no longer listed for new setup. Each one still appears on the Connectors page when you have a saved connection to it, and those connections keep running with their access and permissions. An agent that needs one of those services can still ask you for it with a connection card on its task.
Start here
- Connect your first connector — a worked setup, end to end, finishing with one read-only task an agent actually runs.
- How connector access works — identities, grants, agent selection, and what "effective policy" means.
What you can connect
The Connectors list holds three kinds of connection. They look alike — each one is a credential Paperclip stores for you — but you set them up differently and different people use them.
App integrations let agents work with information and actions inside an app you already use: read a Notion page, open a Linear issue, look at a Sentry error, search a Drive folder. Setup ends in a list of individual actions you switch on or off. The agent starts the work.
Model providers supply the credential Paperclip uses to run models. Setup ends in a stored key or subscription, and there is no action list, because nothing here is a tool an agent can call. Paperclip uses it on the agent's behalf whenever the agent thinks. If you have ever wondered why your Anthropic connection has no permissions to set, this is why.
Chat channels — Experimental let people talk to your agents from a messaging app. An administrator must enable Chat connectors, which is off by default. Setup connects a provider app or bot to Paperclip; eligible messages can start or continue agent tasks.
One service can do more than one of these. Slack offers an app integration and an experimental chat channel, configured separately with separate credentials. Google Chat is an app integration: agents read and post in it, but people cannot use it to talk to an agent. GitHub has two entries: GitHub for repository tools under Developer tools, and GitHub Code Review Bot for the experimental review bot under chat channels. The review bot has its own experimental setting, GitHub review bots, separate from Chat connectors.
Connection methods
Every row in the catalog lists the methods that connector supports. The names mean the same thing everywhere:
| Method | What it means |
|---|---|
| Connect with Paperclip | Paperclip holds the OAuth client and you approve access in the provider's own consent screen. Offered only on instances enrolled with Paperclip Cloud. |
| Sign in with the provider | Paperclip registers itself with the provider's authorization server at connect time, then you approve access there. Nothing to create first. |
| Your own OAuth app | You register an OAuth client with the provider and add Paperclip's callback URI before connecting. Use this when the provider does not support automatic registration, or when you want the client to be yours. |
| API key | You generate a key or token in the provider's console and paste it into Paperclip. |
| Provider app registration | You create an app or bot in the provider's developer console, give it the documented permissions and events, and connect its credentials. Provider-specific channel guides describe exceptions, such as managed inboxes. |
| No credential | Nothing to sign in to. You identify the target — a store domain, a generated URL, a shared file — and that is the whole setup. |
Which methods you actually see depends on your instance. Connect with Paperclip is hidden unless your instance is enrolled with Paperclip Cloud and Cloud advertises a profile for that connector. Every chat channel is behind the Chat connectors instance setting, which is off by default, except the GitHub Code Review Bot, which has its own GitHub review bots setting, also off by default. AgentMail is not: its email inboxes are available on every instance. The memory providers are behind a separate Memory connectors setting, also off by default. Each connector page lists its own prerequisites.
What setup looks like
Connecting a catalog app is one screen. Paperclip picks the recommended method for you and asks only for what proves who you are: a provider sign-in, a key, or an address. A catalog card that says Add key means you will paste a key; every other card says Connect.
Above the main button, one line states the access Paperclip will use, such as "Connects for everyone in your organization, available to all agents." When an agent asked for the connection from a task, it reads "available to the agent that asked for it" instead. Select Change to pick a different identity, narrow the agents, or choose another method or capability; everything there can also be changed afterwards. Individual actions are set on the Permissions tab once you are connected — see Set action permissions.
The catalog
Find an app and open its setup guide.
Google Workspace
Start with Google Workspace for a new Google account. The single-service connectors below it are listed for connections you already have; they no longer appear on the Connectors page for new setup.
| Connector | What you can do | Connection methods |
|---|---|---|
| Google Workspace | One Google account for Gmail, Drive (including Docs, Sheets, and Slides), and Calendar. You choose which services to connect and whether each is read-only or read and write before you sign in, and agents get only the actions you picked and Google granted. This is the Google connector offered for new setup. | Connect with Paperclip · Your own OAuth app |
| Gmail | Google's email service. Agents search and read mail; on a draft connection they create drafts for you to review, and on an opt-in archive connection they remove the Inbox label from threads. Sending, deleting, and other label changes are never enabled. | Connect with Paperclip · Your own OAuth app |
| Google Calendar | Google's calendar. Agents read calendars, and on a write connection create and change events. | Connect with Paperclip · Your own OAuth app |
| Google Chat | Google Workspace's team messaging. Agents search and read conversations, and on a write connection post messages. This is an agent tool, not a way for people to talk to an agent. | Connect with Paperclip · Your own OAuth app |
| Google Docs | Google's word processor. Agents read documents, and on a write connection edit them. | Connect with Paperclip · Your own OAuth app |
| Google Drive | Google's file storage. Agents search and read files, and on a write connection create and copy them. | Connect with Paperclip · Your own OAuth app |
| Google People | The contacts and directory behind a Google account. Agents look up people and profiles. Read-only; there is no write connection. | Connect with Paperclip · Your own OAuth app |
| Google Sheets | Google's spreadsheets. Agents read sheets, and on a write connection update them. A third path shares individual spreadsheets with a Paperclip robot account instead of connecting a Google identity at all. | Connect with Paperclip · Your own OAuth app · No credential |
| Google Slides | Google's presentations. Agents read decks, and on a write connection edit them. | Connect with Paperclip · Your own OAuth app |
| Google Workspace Search | One read-only search that spans Gmail, Drive, Calendar, and Chat in a single Google account. Use it when an agent needs to find something without knowing which app holds it. | Connect with Paperclip · Your own OAuth app |
Note: Google app verification is pending. See Google setup requirements for unverified-app warnings, account restrictions, and the separate Developer Preview prerequisites.
Productivity and collaboration
| Connector | What you can do | Connection methods |
|---|---|---|
| Arcade | Tool gateway combining the tools you select from several servers behind one URL. Agents use the exposed tools; authentication follows your gateway settings. Setup is unverified. | Gateway URL · Browser sign-in or token/headers |
| Asana | Work management for team projects, tasks, and goals. Agents work with the projects and tasks the authorizing Asana account can reach. | Connect with Paperclip · Your own OAuth app |
| Calendly | Scheduling links and bookings. Agents review event types and availability, and where you allow it, use Calendly's scheduling actions. The sign-in grants read and write together, so set booking actions to Ask first. | Sign in with Calendly |
| Composio | App discovery and execution through Composio Connect or a configured session. Upstream app authorization happens in Composio; broad execution actions need separate review. Setup is unverified. | Composio Connect sign-in · Configured session URL and headers |
| Executor | One remote endpoint for configured integrations. Agents use its exposed actions within Executor's upstream policies. Setup is unverified. | Deployment URL · Browser sign-in or token/headers |
| Fireflies — Live exception; not yet stable | Meeting transcripts, summaries, and action items the connected account can access. Setup is unverified. | Sign in with Fireflies · API key |
| Jira | Atlassian's issue tracker for software teams. Agents work with the issues on the Jira site you connect. | Sign in with Jira |
| Linear | Issue tracking for product and engineering teams. Agents create, update, and read issues across whatever the authorizing Linear account can reach. A Read-only API key gives a read-only connection. | Sign in with Linear · API key |
| Make | Automation scenarios you expose to MCP in Make. Agents look up organizations and teams and run scenarios; set scenario runs to Ask first. | Sign in with Make · MCP token |
| Miro | Shared online whiteboards for diagrams, planning, and workshops. Agents work with the boards your Miro account can reach. | Sign in with Miro |
| Notion | Workspace for notes, documents, and databases. Agents read and update content accessible to the Notion account you authorize. Paperclip does not add a page-level filter. | Sign in with Notion |
| Todoist | Task lists for individuals and small teams. Agents work with the tasks and projects in your Todoist account. | Sign in with Todoist |
| Zapier | Automation service that wires thousands of apps together. You choose the actions in Zapier and paste one generated URL, so the agent gets the actions you put in it. Treat that URL as a secret — its token is part of the address. | Generated MCP URL |
MCP aggregators
Services that connect to many apps for you and hand them to Paperclip through one MCP server. Use one for an app that has no native connector here. See Connect apps through an MCP aggregator for setup and for how an agent asks you to pick one.
| Connector | What you can do | Connection methods |
|---|---|---|
| Arcade | Use the tools exposed by an Arcade MCP gateway you set up. | Connect MCP server |
| Composio | Discover and use connected apps through Composio Connect. | Composio Connect |
| Executor | Use the tools exposed by your Executor workspace. | Connect MCP server |
| Zapier | The actions you put on a Zapier MCP server. Also listed under Productivity. | Generated MCP URL |
Communication
| Connector | What you can do | Connection methods |
|---|---|---|
| AgentMail | Email inboxes built for software agents. Gives an agent its own inbox and turns each email conversation into a Paperclip task. Available on every instance; no experimental setting needed. | API key |
| Resend | Transactional email delivery for developers. Agents work with the sending domains and delivery records of the Resend team you sign in to. The grant is full access, including sending and deletion, so set actions deliberately. | Sign in with Resend |
| Slack | Team messaging. As an app integration, agents read and post within the authorizing account's access. Its separate experimental chat channel lets people work with one agent from Slack. | Agent-tool setup: see compatibility notice · Provider app registration for chat |
Developer tools
| Connector | What you can do | Connection methods |
|---|---|---|
| Braintrust | AI evaluation and observability. Agents query traces, manage datasets and scorers, and run evaluations, in the US or EU region. Dataset row edits and automation activation are destructive, and evaluations can use credits. | Sign in with Braintrust · API key |
| Browser Use Cloud | Run browser tasks in Browser Use Cloud, open a live browser session from the task, and reuse a browser profile. | API key |
| Cloudflare | DNS, CDN, and edge compute. Agents work with the account your Cloudflare sign-in can reach. | Sign in with Cloudflare · API key |
| Context7 | Up-to-date public library documentation and code examples. Read-only and keyless; anonymous rate limits apply, and private repositories are out of reach. | No credential |
| Daytona | Manage your own Daytona account: sandboxes, files, commands, Git, snapshots, and volumes, across 274 actions. Separate from the Daytona sandbox provider that runs agents. Gate deletions and commands. | API key |
| GitHub | Code hosting and review. Agents use repository tools with the connected credential's permissions. See the guide for setup and shell-access boundaries. The pull-request review bot is its own entry, GitHub Code Review Bot, under Chat channels. | Connect with Paperclip · API key |
| Netlify | Hosting and deploys for web front ends. Agents work with the teams and sites your Netlify account can reach. | Sign in with Netlify |
| PagerDuty | On-call scheduling and incident response. Agents work with the incidents, services, and schedules your token can reach. Choose the US or EU service region when you connect. | API key |
| Postman | API development and testing workspace. Agents work with the collections and APIs your Postman account can reach. Choose a toolset: Minimal, Code, or Full. Minimal is not read-only; set action permissions separately. | Sign in with Postman · API key |
| Railway | Infrastructure work in consented workspaces, including service, deployment, and bounded log actions when API access is accepted. Live qualification is pending; container commands require separate SSH setup. | Sign in with Railway |
| Sentry | Error and performance monitoring. Agents investigate errors, releases, and production issues in the Sentry organization you authorize. | Sign in with Sentry |
| Superagent | Security platform. Agents review security findings, start red-team reports, and score content and packages. Some actions use credits or delete permanently, so gate them. | API key |
| Unikraft | Manage your Unikraft Cloud organization: instances, services, autoscaling, volumes, images, certificates, and sandbox commands and files, across 105 actions in five metros. Gate deletions and sandbox commands. | API key |
Data and analytics
| Connector | What you can do | Connection methods |
|---|---|---|
| Airtable | Spreadsheet-database hybrid for structured team data. Agents work with the bases your Airtable sign-in or personal access token can reach. | Sign in with Airtable · Personal access token |
| Bitly | Link shortening and click analytics. Agents read links, QR codes, and click data. The grant covers Bitly's full tool list, including permanent link deletion and Sites publishing, so keep writes and deletes Off. | Sign in with Bitly · API key |
| ClickHouse | Columnar database built for analytical queries over very large datasets. Agents query the ClickHouse Cloud service you name. | Sign in with ClickHouse |
| Enterpret | Customer feedback analysis. Agents ask about feedback themes and pull verbatim quotes with citations, through Enterpret's read-only server. Graph queries start on Ask first. | API key (auth token) · Sign in with Enterpret |
| Gauge | AI-answer visibility and content workflows. Agents read brand mentions, SEO, and traffic data for one Gauge organization. Content tools can publish live to your CMS; set publish actions to Ask first. | Sign in with Gauge · API key |
| GSC Wizard | Search performance from Google Search Console, Bing Webmaster, and analytics for the properties you connected in GSC Wizard. Some tools submit URLs for indexing; review those separately. | Sign in with GSC Wizard · MCP key |
| Mixpanel | Product analytics for user and event behaviour. Agents work with the events and reports your Mixpanel account can reach. | Sign in with Mixpanel |
| Neon | Serverless Postgres with branching. Agents manage projects and branches, run SQL, and inspect schemas, optionally pinned to one project. Use a development project. | Sign in with Neon · API key |
| PostHog | Product analytics with session replay, feature flags, and experiments. Agents analyse product usage, errors, flags, and experiments. | Sign in with PostHog · API key |
| Supabase | Hosted Postgres with authentication, storage, and edge functions. Agents work with the Supabase project you scope the connection to. Use a development project. | Sign in with Supabase · API key |
| Windsor.ai | Marketing, analytics, CRM, and warehouse data across the sources you linked in Windsor.ai. Sign-in always includes the provider's delete scope, and some tools change campaigns or budgets. | Sign in with Windsor.ai · API key |
Google Sheets is the other common home for tabular data; it sits under Google Workspace with the rest of the suite.
Content and design
| Connector | What you can do | Connection methods |
|---|---|---|
| Box | Enterprise file storage and sharing. Agents work with the files and folders your Box integration can reach. | Your own OAuth app |
| Cloudinary | Image and video hosting with on-the-fly transformation. Agents work with the assets your Cloudinary roles allow. | Sign in with Cloudinary |
| Ditto | Product copy and style guides. Agents search approved text, check wording against your style guide, and create, edit, and translate text. Deletes are permanent, so set them to Ask first. | Sign in with Ditto · API key |
| Webflow | Visual website builder with a CMS behind it. Agents work with the sites and collections your Webflow roles allow. | Sign in with Webflow |
| Wix | Website builder and hosting. Agents work with the sites your Wix account can reach. | Sign in with Wix |
Commerce and finance
| Connector | What you can do | Connection methods |
|---|---|---|
| Shopify | E-commerce platform for online stores. Agents search a store's products and policies and manage shopping carts. You give a store domain; there is no sign-in, and no access to the store's admin. | No credential |
| Stripe | Payments, invoicing, and billing. Agents read customers, invoices, and payouts in the account you connect. | Sign in with Stripe · API key |
AI tools
Services whose subject matter is machine learning, plus the web search and page-reading services agents use to look things up. They are ordinary app integrations with action lists, which is what separates them from the model providers below.
| Connector | What you can do | Connection methods |
|---|---|---|
| Exa | Public web search and page reading through Exa's keyless server. Two read-only actions; no account or key. | No credential |
| Firecrawl | Search, scrape, and parse public web pages. Keyless access offers three read actions; an account adds crawl and research jobs that spend team credits. | Try keyless access · Sign in with Firecrawl · API key |
| Hugging Face | Public hub for open machine-learning models, datasets, and demos. Agents search models, datasets, and documentation. The public Hub method needs no account or key, which makes it the shortest connector to try first; sign-in adds private repositories plus repository and job access, so switch its writes off while you learn. | No credential (public Hub) · Sign in with Hugging Face |
| Parallel Search | Public web search and page extraction through Parallel's keyless server. Two read-only actions; no account or key. | No credential |
| Tavily | Web search and page extraction using your Tavily account's credits. Read-only. Sign-in needs a public HTTPS Paperclip address. | Sign in with Tavily |
| Telem.AI | Web search and page reading across many search providers with one API key. Optional routing, tier, and provider lists. Usage bills to your Telem account. | API key |
| You.com | Hosted web intelligence. The free profile has a reduced read-only tool set; provider limits apply. Setup is unverified. | Sign in with You.com · API key · No credential (free profile) |
Memory — Experimental
Long-term memory services that agents can store to and recall from across runs. Setup for all of them is hidden until an administrator turns on Memory connectors in the instance's experimental settings; the setting is off by default. Turning it off again hides setup only — existing connections keep running.
| Connector | What you can do | Connection methods |
|---|---|---|
| Mem0 | Hosted long-term memory for AI applications. Agents store and recall facts across runs in your Mem0 project. | API key |
| Zep | Retrieve temporal graph memory and authorized business context. Needs a Zep project with Memory MCP enabled; project API keys do not work here. | Sign in with Zep |
| Supermemory | Search and save shared memories, documents, and profiles. You choose which spaces the connection may use. | Sign in with Supermemory |
| Cognee | Build and recall shared graph memory from documents and conversations, using your Cognee Cloud tenant. It runs Cognee's MCP client on Paperclip's runtime host, so a public deployment needs a trusted MCP runtime host. | API key |
| Honcho | Remember conversations and retrieve context about peers. | API key |
See Memory connectors for the experimental gate, provider differences, and what is shared through a connection.
Model providers
These hold the credential Paperclip uses to run models. They publish no actions and appear in no permission list. See Agent adapters for which adapter uses which provider.
| Connector | What you can do | Connection methods |
|---|---|---|
| Anthropic | Run Claude models. A subscription signs in with your Claude plan; an API key can be scoped and rotated separately. | Claude subscription · API key |
| OpenAI | Run OpenAI models. | OpenAI subscription · API key |
| OpenRouter | Run models from many vendors through one credential and one bill. | API key |
| GitHub Copilot | Run agents on your Copilot plan through Paperclip Runner. Model access only; set it up with Connect Copilot on the GitHub row. | Personal access token |
| Grok | Run xAI's Grok models. | Grok subscription · API key |
| Custom model providers | Use a compatible custom or local endpoint through a supported agent harness. Compatibility depends on the API route. | API key · Endpoint URL |
Check AI usage explains provider-supported usage snapshots and reconnect messages. These credentials are separate from assistant connections (MCP), which let an external assistant sign in as a person and work with Paperclip.
Chat channels — Experimental
These give people a place to talk to an agent. They require an administrator to enable Chat connectors in the instance's experimental settings; the setting is off by default. The GitHub Code Review Bot is the exception: it has its own GitHub review bots setting, also off by default. Setup registers an app or bot with the provider and points it at a Paperclip webhook; one connection carries one agent.
| Connector | What you can do | Connection methods |
|---|---|---|
| GitHub Code Review Bot | Have an agent review pull requests and answer mentions in GitHub under its own App identity. Choose the agent, then GitHub creates and installs the App for you. Needs the separate GitHub review bots setting. | Provider app registration |
| Discord | Mention the agent in a server channel and Paperclip opens a thread, keeping it tied to one task. | Provider app registration |
| iMessage Photon | Message the agent from Apple Messages through a Photon Cloud project. Shared Pro lines carry direct messages; a dedicated line also carries groups you enable. | Provider app registration |
| Microsoft Teams | Message the agent in a chat, a team channel, or a group chat. Needs a work or school Microsoft 365 organization; personal Teams accounts cannot complete the setup. | Provider app registration |
| Slack | Message the agent directly, or mention it in a channel to start a task from that thread. Set up separately from the Slack agent tool on the same page. | Provider app registration |
| Speko | Talk to an agent by voice from a task, and optionally by phone. | Provider API key |
| Telegram | Message the agent through a bot you create with BotFather. | Provider app registration |
Can't find your app?
The catalog above describes the documented connectors for this release. If the service you want is not in it, these are the paths that exist.
Check your own instance first. Your installed version may offer a different catalog, and the methods it offers you depend on whether the instance is enrolled with Paperclip Cloud and whether chat channels are switched on. What Connectors shows you is authoritative; this page is not.
Connect it as a custom server. If the service publishes its own remote MCP server, you can connect it directly without waiting for a catalog entry. It gets the same identity, grant, agent-selection, and per-action controls as a catalog connector. See Connect a custom MCP server for the compatibility requirements and the two ways to supply a URL.
Reach it through an MCP aggregator. Arcade, Composio, Executor, and Zapier each connect to many apps on your behalf. You do not always have to go looking: when an agent needs an app with no native connector, it asks you which of these providers to use, or None for now. With Zapier you pick the actions on Zapier's side, which also means the agent cannot exceed them.
Give the agent a workspace instead. Some work does not need a connector at all. Anything an agent can do with a shell and a CLI belongs in an execution workspace — see Connect an agent to a GitHub repo for the pattern, and Add an MCP server to an agent for the adapter-level route.
If the provider gates access. A few providers review and approve MCP clients before an independently registered one will work. Nothing you do in Paperclip changes that, and the approval has to come from the provider. Providers Paperclip recognizes but does not list names the ones in this snapshot and what each is waiting on.
What a connection does not grant
Two things surprise people, and both are worth knowing before you connect anything.
A connector is not a permission. Connecting Notion does not let every agent edit your Notion. The connection holds the credential; a separate set of choices decides which humans it belongs to, which agents may use it, and which individual actions those agents may call. How connector access works is the short version, and Set action permissions is the control itself.
Most action lists come from the provider, not from Paperclip. For a connector that talks to a provider-hosted server, the provider publishes the tools. Paperclip reads that list, classifies each entry as read, write, or destructive, and lets you set each one to Allowed, Ask first, or Off. When the provider changes its server, your list changes with it — so the authoritative list of what a connector can do is the one in your own instance, not the one on its page here.
Shared guides
Use these guides for the controls your connector exposes. Action permissions and review requests apply to agent-tool connections; model credentials and chat channels have different controls.
| Guide | Use it when |
|---|---|
| Share a connector with people and agents | Deciding who the credential belongs to and which agents may use it. |
| Use separate accounts for people and agents | One agent needs its own account rather than borrowing a person's. |
| Set action permissions | Turning individual actions to Allowed, Ask first, or Off. |
| Answer a connector review request | An agent is waiting on your OK for an Ask-first action. |
| Verify a connector and fix a broken one | Checking a new connection, or a working one has stopped. |
| Reauthorize, revoke, or disconnect | Rotating a credential, removing one person's access, or deleting the connection. |
| Connect a custom MCP server | The service is not in the catalog, or you have a URL or config to paste. |
| Connect apps through an MCP aggregator | The app has no native connector, or an agent asked you to pick an external provider. |
| Connection instructions | Saving service-specific instructions an agent should receive with a connection. |
Related
- Tool Gateway — what sits between an agent's tool call and the outside service.
- Providers Paperclip recognizes but does not list — the withheld and provider-gated inventory in this snapshot.
- Add an MCP server to an agent — the adapter-level path, which is separate from connectors.
- Connect an agent to a GitHub repo — git and
ghinside a workspace, which is not the same thing as the GitHub connector. - Connections v3 (Apps) — the storage and authorization foundation these connectors are built on.