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]>
6.5 KiB
paperclip_version, seo_title, seo_description
| paperclip_version | seo_title | seo_description |
|---|---|---|
| v2026.1106.0 | Connect a Custom MCP Server | Add an MCP server that is not in the Paperclip catalog by URL, by pasting a config, or with a provider-generated URL, and govern it like any connector. |
Connect a custom MCP server
The catalog covers the providers Paperclip has reviewed. For anything else — a server you run, a provider that is not listed, or a URL a provider generated for you — there are three entry points on the Connectors page.
Connect your own MCP server
"Enter the URL for a custom or self-hosted MCP server."
Select Connect your own MCP server, paste the full http or https address, and let Paperclip work out what the server needs: "Paperclip asks the server what it needs and walks you through it. Start here."
What happens next depends on the server:
- It advertises OAuth. Paperclip registers a client and takes you through browser sign-in. If the authorization server supports neither Client ID Metadata Documents nor dynamic registration, Paperclip stops and asks you for a client you registered yourself rather than losing the draft connection.
- It wants a key. Paperclip asks for it, and states where it goes: "Paperclip sends your key as an Authorization header." Custom headers are available for servers that expect something else.
- It wants nothing. Paperclip says so plainly: "The server is open to anyone with the address." Treat that address as the credential.
Staying signed in after the access token expires
Many OAuth servers hand out short-lived access tokens. To keep a custom server connected past that, Paperclip also asks for offline_access when the server's authorization server lists it as supported and does not rule out refresh tokens, even if the MCP server itself only mentions its tool scopes. When the provider issues a refresh token, Paperclip renews access in the background and agents keep working without another sign-in.
The provider still decides whether to issue one. If it does not, the connection needs a fresh sign-in each time the access token expires. An older custom connection made before this change picks up refresh support the next time you select Reconnect.
A server outside the reviewed catalog is labelled Unverified server. That label is a statement about review, not about whether the connection works.
Paste a config
"Paste an existing setup snippet and connect it."
If you already have a working MCP client configuration — from another tool, a provider's quickstart, a teammate — Paste a config reads the server URL and credential shape out of it instead of making you retype them. Paperclip validates the result before saving.
Paste a provider-generated URL
Some providers hand you a single URL with the token already embedded. Zapier is the catalog entry for this shape: "Paste the complete MCP URL Zapier gives you, including its token."
The URL is a secret. Anyone holding it holds the access. Paperclip stores it as one, and the connection's action permissions apply the same way they do to any other connector.
Arcade, Composio, and Executor work much the same way — you paste the MCP URL their service gives you, and sign in if it asks — but each has its own catalog entry with setup steps for that provider. If the app you want is reachable through one of them, start with Connect apps through an MCP aggregator instead of a custom server.
The two generic definitions behind this
Paperclip has two catalog entries that exist to back the generic paths rather than to be browsed:
| Definition | Slug | Shape |
|---|---|---|
| OAuth app | oauth-generic |
"Connect a provider using your own OAuth client." You supply a client ID and client secret; dynamic registration is used when the provider supports it. |
| API key app | api-key-generic |
"Connect an API using a key from your provider." You supply a key, which Paperclip presents on each call. |
Neither appears in the Connectors list. They are reached through the generic entry points above.
Governance is identical
A custom server is not a lesser citizen. The same four gates apply: the connection, the identity, agent access, and per-action permission. The action list is read from the server you pointed at, classified read, write, or destructive, and each entry set to Allowed, Ask first, or Off.
Because the server has not been reviewed, two habits are worth keeping:
- Start with every write Off and promote deliberately. A name-based classification is the fallback when a server publishes no annotations, and an unfamiliar naming scheme can under-classify.
- Re-run Refresh actions after you change the server, then review the list. On a pasted-URL connection a newly discovered action becomes active under the policies already in force — it is not held back for approval — so a refresh can widen what agents can call.
Add saved guidance
A custom connection can store optional agentInstructions through the connection API. The guidance is included only for authorized runs that can use at least one of its tools, and it never grants additional access.
The current Agent instructions editor appears only when a connector declares a template; it is not shown on every custom server. Paperclip does not automatically adopt the server's initialization prose as trusted instructions. See Saved connection instructions for the setting and its behavior.
Not the same as an adapter MCP server
Attaching an MCP server to an agent's runtime — through the adapter's own configuration — is a different mechanism with different governance. Paperclip's action permissions and review queue do not sit in front of it. See Add an MCP server to an agent for that path, and pick it deliberately rather than by accident.
Related
- Connectors
- How connector access works
- Set action permissions
- Zapier
- Connect apps through an MCP aggregator
- Tool Gateway
Sources
- Connection setup — one-screen access choices and setup behavior.
- Remote MCP setup — provider-specific connection controls.