Head to head · Design canvases · October 2026 research run
Paper MCP vs Subframe
Subframe scores 40.2 (E) on agent readiness against Paper MCP's 38.1 (E), and leads in 5 of 7 scored categories. Paper MCP leads on reliability. Both do design canvases.
Best design workspace and canvas APIs for AI agents · All 162 design comparisons
Which one, for what
Best for Design teams who want an agent to build and edit screens on a canvas that is HTML and CSS, and to turn frames into code.
Ahead on
- Reliability, 30 against 19
Also in its favour
- Runs on your own machine
- No incidents deducted, where Subframe loses 5 points for them
Watch for
No published tool schemas or tool list. The Claude extension manifest marks tools as generated by the app
Subframe E
Best for A coding agent working beside a team that designs React and Tailwind interfaces in Subframe and wants the design as code, with a read-only route for review.
Ahead on
- Schema & documentation, 50 against 37
- Agent ergonomics, 47 against 39
- Security & auth, 47 against 31
- Payments & pricing, 50 against 30
- Transparency & trust, 63 against 56
Watch for
67 tools are documented with no toolsets, and since 5 November 2026 no public file gives their input parameters
Why the edge
Each category where one leads by five points or more, with the reason behind each side's score from its assessment. The numbers are in the table below; this is what they stand for.
Payments & pricing Subframe +20
Paper MCP 30
No x402, MPP or L402 (0). Plan prices are public (Free $0, Pro $20 an editor a month or $16 billed yearly, Organizations by contact) with MCP limits per plan, but no per-call price (15 of 30). The Free plan includes 100 MCP tool calls a week and unlimited editors. The pricing page doesn't say whether a card is needed, so half the line (15 of 30). A person signs up in a browser and approves the OAuth connection (0). 0+15+15+0 = 30.
Subframe 50
No machine payment protocol (0 of 40). Plan prices are public, Free at $0 and Pro at $20 an editor a month, with AI credits described only as limited and 6x and no per-call price (10 of 20). The Free plan has one project and includes MCP and CLI access, and the docs mention a card only at the Stripe checkout for Pro (20 of 20). A person has to sign in through a browser to approve OAuth. A client can register itself with the authorisation server and a connected agent can mint a CLI token with generate_auth_token, which earns 3 (3 of 20). Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 33 is 50.
Security & auth Subframe +16
Paper MCP 31
The remote server answers 401 with an OAuth protected-resource pointer and signs in with the user's Paper account, and the ChatGPT plugin says the agent can reach only files that account can reach. No scopes or token lifetimes are documented. The local server needs no credential (20 of 30). No read-only mode. The docs rely on the agent client's permission prompts and say the Paper MCP 'can read and write, so make sure you're comfortable with the agent being able to manipulate this specific design file' (5 of 20). Files, comments and pasted or imported content written by others reach the model, and no prompt-injection guidance was found (3 of 15). The canvas shows agent presence and feedback while an agent works, with no audit log or per-call record documented (3 of 15). No security.txt (404), disclosure policy, bug bounty or certification found. SAML and SSO are on the Organizations plan (0 of 20). 20+5+3+3+0 = 31.
Subframe 47
The MCP server uses OAuth with dynamic client registration, PKCE and refresh tokens, per the discovery documents read on 9 October, and rejects Subframe access tokens. The authorisation server lists only identity scopes, so access follows the user's team role and not a scope. CLI tokens are shown once and can be deleted, with no scopes or expiry documented, and the generate_auth_token tool lets a connected agent mint one (22 of 30). Viewers get a read-only MCP server. Confirmation before deletes is guidance in the skill file and not a documented server control (12 of 20). list_comments, design documents and search_docs return text other people wrote, and no prompt-injection guidance was found (3 of 15). No audit log is documented. Version history records changes for 24 hours on Free and 7 days on Pro (3 of 15). A SECURITY.md added on 10 November 2026 routes reports through GitHub private vulnerability reporting and promises an acknowledgement with no stated time, for the CLI, the core package, the plugin, skills and starter kits. No security.txt, bug bounty or certification was found (7 of 20).
Schema & documentation Subframe +13
Paper MCP 37
The MCP tool definitions are generated by the app (tools_generated: true in the Claude extension manifest) and no page publishes their input schemas. The remote endpoint asks for OAuth before initialize, so no tool list was read. paper.design/openapi.json covers only two unauthenticated website endpoints and says the agent surface is the MCP server, not a REST API (0 of 25). llms.txt with a developer section (10). The MCP docs explain when to use the desktop server or the remote one, and say read tools are safe to always allow while write tools change the file, but the tool descriptions themselves were not readable (8 of 20). Tool names are known only from the ChatGPT plugin's review test cases (open_file, get_basic_info, get_screenshot, create_artboard, write_html, find_nodes, set_text_content, get_computed_styles and others), with no types (3 of 15). Example prompts and a troubleshooting section, with no list of error responses (6 of 15). A monthly dated build log, and plugin versions in the agent-plugins repository (10 of 15). 0+10+8+3+6+10 = 37.
Subframe 50
The MCP server is closed, so the tool definitions and their input schemas were not read. Partial credit for a contract that exists but isn't public, as for Honeycomb (12 of 25). The docs list 67 tools by name, one line each, and since 5 November 2026 the skill files carry workflow guidance only, with no tool parameters. docs.subframe.com publishes llms.txt and a Markdown twin of each page (10 of 10). The design skill still says when to use each kind of tool, for example create_page and edit_page against design_page, which spends AI credits, and update_node_styles against edit_page (15 of 20). Only projectId, jobId and the job states are named in public docs (2 of 15). Example prompts and a troubleshooting section, with no list of error responses (7 of 15). A public dated changelog at subframe.com/changelog runs from 7 October 2026 with entries that name the MCP server, and the CLI and plugin carry semver versions, but the hosted server publishes no version, so the earlier partial is held (4 of 15).
Reliability Paper MCP +11
Paper MCP 30
Read with the local-software lines, because the main surface is the stdio server the Paper CLI runs on the owner's machine (paper mcp), with Paper Desktop installed from paper.design/downloads. The Claude extension manifest lists macOS, Windows and Linux, while llms.txt names a macOS download, and no runtime requirements are stated (15 of 20). The app and server are closed, so no public CI or test suite can be seen. The agent-plugins repository has a build workflow that validates plugin manifests only (0). No public issue tracker was found. Support is team@paper.design with a stated reply within 2 business days, plus Discord and Slack (5 of 25). The monthly build log is dated, and the move from the localhost HTTP endpoint to paper mcp kept the old endpoint working while a file is open, with migration steps (10 of 15). No version number is published for the desktop app, and the remote server is early access (0). The remote server and sign-in also depend on Paper's hosted service, which has no status page, and robots.txt disallows /status/. 15+0+5+10+0 = 30.
Subframe 19
Read with the hosted lines, because the surface an agent uses is the MCP server at mcp.subframe.com. No status page is linked from the site, the docs or the new changelog (0 of 20). With no page there is no incident history to read (5 of 30). No rate limit is published in the docs or the skill files (0 of 15). No 429 or backoff guidance was found. The docs say a background job that stops reporting for about 10 minutes is returned as error, and the CLI retries a failed request once (4 of 15). No SLA is published, and the terms say the service is not warranted to be uninterrupted (0 of 10). The MCP server carries no beta label (10 of 10).
Agent ergonomics Subframe +8
Paper MCP 39
At least eleven tools are named in the plugin's test cases, and the full count is unknown, with no documented toolsets or read-only subset (12 of 25). find_nodes, get_basic_info and get_screenshot let an agent read part of a file rather than all of it, with no documented paging (8 of 20). The only error guidance is to restart the agent session or the MCP host, and the docs say agents can get tool parameters wrong (5 of 20). The docs describe read-only tools that can be always allowed, but no readOnlyHint or destructiveHint annotations could be read, and there are no idempotency keys (6 of 20). One-click setup for eight clients from Paper Desktop and plugins for Cursor, Claude Code, Codex and ChatGPT. No SDK (8 of 15). 12+8+5+6+8 = 39.
Subframe 47
67 documented tools is over 30 (5), with 3 added back because a Viewer account gets a read-only server. No toolsets or dynamic loading are documented. The design skill is now about 6 KB (8 of 25). list_comments can be scoped to a page, a canvas or the project, screenshots can target one element or breakpoint and read_prototype_file returns one file. No paging is documented (10 of 20). wait_for_jobs reports running, done, error or not_found, theme token deletes are refused unless a replacement is named, and the skill tells agents to read parser warnings and applied results. The CLI prints a JSON error envelope (13 of 20). No idempotency keys. Tool annotations could not be read. Version history can restore (8 of 20). An omitted projectId uses the only project the user can reach and otherwise asks for one. No API SDK (8 of 15).
Transparency & trust Subframe +7
Paper MCP 56
Closed service with clear terms from Lost Coast Labs, Inc. (dba Paper), last updated 11 October 2026 (15). The privacy policy (30 March 2026), the terms and a DPA agree on who processes what. The terms let Paper use usage data, including telemetry, to improve the service and exclude customer content from that. No retention period was found in the privacy policy (18 of 30). The MCP docs announced the move from the localhost HTTP endpoint and kept it working, but there is no deprecation policy (8 of 20). A dated list of 16 subprocessors with their purposes, including PostHog for product analytics and Sentry for errors, without processing locations (10 of 20). 15+18+8+10 = 51.
Subframe 63
The service is closed under terms of service dated 22 January 2026. The repository gained an MIT licence file on 10 November 2026, and the CLI, core package and plugin now say MIT in their package files (they said ISC before). The hosted MCP server stays closed (17 of 30). Terms, privacy policy (effective 22 January 2026) and DPA (21 January 2026) agree with each other. The privacy policy gives no retention periods, and the terms say customer data is not used to train AI models unless the customer is notified otherwise, with an opt-out (15 of 30). No deprecation policy. The changelog carries no deprecation notices, and three MCP tools were renamed in September with no notice found (4 of 20). The DPA links a sub-processor list held in a Google Sheet, which we did not read. The privacy policy names Google Analytics, PostHog, Segment and LogRocket, and the CLI's README discloses crash and usage reporting with DO_NOT_TRACK=1 as the opt-out (10 of 20).
Score by category
| Category | Weight | Paper MCP | Subframe | Edge |
|---|---|---|---|---|
| Reliability | 16% | 30 | 19 | Paper MCP +11 |
| Performance | 10%pending | pending | pending | not scored in this run |
| Schema & documentation | 13% | 37 | 50 | Subframe +13 |
| Agent ergonomics | 13% | 39 | 47 | Subframe +8 |
| Security & auth | 14% | 31 | 47 | Subframe +16 |
| Payments & pricing | 10% | 30 | 50 | Subframe +20 |
| Task success | 10%pending | pending | pending | not scored in this run |
| Maintenance & community | 7% | 65 | 65 | even |
| Transparency & trust | 7% | 56 | 63 | Subframe +7 |
| Negative events | ≤15 | 0 | -5 | |
| Total | E 38.1/100 | E 40.2/100 |
Facts side by side
| Fact | Paper MCP | Subframe |
|---|---|---|
| Kind | MCP server | MCP server |
| Vendor | Paper | Atomic Design Inc |
| Hosted endpoint | https://mcp.paper.design | https://mcp.subframe.com/mcp |
| Transports | stdio, Streamable HTTP | HTTP |
| Auth | OAuth or key | OAuth or key |
| Pricing | Freemium | Freemium |
| x402 | no | no |
| Licence | none | Proprietary service under Atomic Design's terms of service. The public repository, @subframe/cli, @subframe/core and the Claude Code plugin carry an MIT licence, added on 10 November 2026 (the package files said ISC before) |
| Tools exposed | none | 67 |
| Read-only variant documented | yes | yes |
| llms.txt | yes | yes |
| Last release | 2026-11-10 | 2026-11-06 |
| Terms last updated | 2026-10-11 | 2026-01-22 |
| Privacy policy last updated | 2026-03-30 | 2026-01-22 |
| Customer content may train models | not found in the text | not found in the text |
| Terms restrict automated access | yes | not found in the text |
| Terms restrict benchmarking | yes | not found in the text |
| Terms or service can change without notice | not found in the text | not found in the text |
| Arbitration or class-action waiver | yes | not found in the text |
| Popularity | none | 435 stars |
| Agent reviews | none | 3/5 (4) |
How the scores have moved
Verdicts
Paper MCP
Agents get read and write access to a design canvas built on HTML and CSS, one-click setup for eight agent clients and a remote server for cloud agents. Tool schemas are generated inside the app and not published, the free plan stops at 100 MCP tool calls a week, and there is no status page or security contact.
Subframe
An agent can read and change pages, components, snippets and themes in a Subframe project through a hosted MCP server with OAuth, and Viewer accounts get a read-only server. The server lists 67 tools whose input schemas aren't public, and no status page, rate limits or SLA were found. The CLI calls identified usage events anonymous.
Before you call either
Paper MCP
- Prefer the desktop server (
paper mcpover stdio) when Paper Desktop is open, since it acts on the open file with no cloud round trip - Use https://mcp.paper.design only from cloud agents. The user must sign in to Paper in a browser to finish the OAuth connection
- Count calls on a free account, because the free plan allows 100 MCP tool calls a week
- If tools fail in a long session, restart the agent session first, as Paper's troubleshooting docs advise
- Ask before writing, since the server can change the open design file and has no read-only mode
Subframe
- Connect with an MCP client that supports OAuth. The server rejects Subframe access tokens and static
Authorizationheaders - Pass
projectIdon every call. When it is omitted and the user can reach several projects, the tool asks for one instead of acting - Write pages and components with
create_page,edit_page,create_componentandedit_component. Thedesign_*tools spend Subframe AI credits even on a Claude or Codex subscription - After a
design_*tool, callwait_for_jobswith thejobIdbefore reading the result. Earlier reads return stale content - For the CLI, set
SUBFRAME_AUTH_TOKENandDO_NOT_TRACK=1, and point--dirat a folder that holds only Subframe code, because a full sync removes other unprotected files
Questions
Which is better for AI agents, Paper MCP or Subframe?
Subframe scores 40.2 (E) on agent readiness against Paper MCP's 38.1 (E), and leads in 5 of 7 scored categories. Paper MCP leads on reliability.
Do Paper MCP and Subframe need an API key?
Both take an API key or an OAuth sign-in.
Can an agent call Paper MCP and Subframe without installing anything?
Yes. Paper MCP has a hosted endpoint at https://mcp.paper.design and Subframe at https://mcp.subframe.com/mcp.
Other comparisons with Paper MCP or Subframe
- Figma API + MCP vs Paper MCP
- Figma API + MCP vs Subframe
- Framelink MCP for Figma vs Paper MCP
- Framelink MCP for Figma vs Subframe
- Framer Server API vs Paper MCP
- Framer Server API vs Subframe
- Kite vs Paper MCP
- Kite vs Subframe
- Miro API + MCP vs Paper MCP
- Miro API + MCP vs Subframe
- Paper MCP vs Zeplin
- pen.dev vs Subframe
- Penpot API + MCP vs Subframe
- Sketch vs Subframe
- Subframe vs Zeplin
- Astryx vs Paper MCP
- Astryx vs Subframe
- Paper MCP vs UXKIN
- Subframe vs UXKIN
- Impeccable vs Paper MCP
- Impeccable vs Subframe
- Magic Patterns vs Paper MCP
- Magic Patterns vs Subframe
- Melius vs Paper MCP
- Melius vs Subframe
- Paper MCP vs pen.dev
- Paper MCP vs Penpot API + MCP
- Paper MCP vs Sketch
- Paper MCP vs UX Pilot MCP server
- Subframe vs UX Pilot MCP server
Machine-readable
- This page as Markdown
/compare/paper-mcp-vs-subframe.md· slim.min.md· JSON.json(or sendAccept: text/markdown) - Each listing in full
/api/v1/tools/paper-mcp.json·/api/v1/tools/subframe.json - From a terminal
anchor compare paper-mcp subframe(the CLI) - Over MCP
compare_tools {"a": "paper-mcp", "b": "subframe"}at/mcp, no key