Head to head · Workflow automation · October 2026 research run

Activepieces API + MCP vs Tray.ai API + MCP

Tray.ai API + MCP and Activepieces API + MCP score within a point of each other on agent readiness, 59.4 (C) and 59.3 (C). Activepieces API + MCP leads on agent ergonomics, payments & pricing, maintenance & community and transparency & trust. Both do workflow automation.

Best workflow automation platforms with APIs for AI agents · All 174 workflows comparisons

Which one, for what

Activepieces API + MCP C

Best for A team that wants an MIT-licensed builder it can self-host or embed, with an agent that builds flows through MCP.

Ahead on

  • Agent ergonomics, 70 against 33
  • Payments & pricing, 60 against 30
  • Maintenance & community, 80 against 60
  • Transparency & trust, 77 against 69

Also in its favour

  • Open source
  • Rated higher by the reviewer agents, 3.0 against 2.5 out of 5

Watch for

Docs say API keys come from the platform dashboard in Platform and Enterprise editions, while the pricing page lists API access on every plan

Tray.ai API + MCP C

Best for SaaS vendors embedding integrations for their customers, with an agent that acts per end user.

Ahead on

  • Reliability, 72 against 65

Also in its favour

  • No incidents deducted, where Activepieces API + MCP loses 11 points for them

Watch for

No published per-plan prices; plans are quoted by sales

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.

Agent ergonomics Activepieces API + MCP +37

Activepieces API + MCP 70

47 MCP tools in groups, so 5, plus 10 back because groups can be switched off per project and ap_search_actions finds actions by task description instead of listing pieces (15). Every endpoint uses seek pagination with limit and cursor, and list endpoints take filters (20, was 18). API errors aren't documented in the spec or the API pages, while the MCP tools return actionable error text, such as a not-found message that names the tool to call next, so the line is partly met (8 of 20). All 47 registered tools set both readOnlyHint and destructiveHint, 17 of them with destructiveHint: true, and 42 set idempotentHint (20, was 17). Few required parameters, but the only official client is the JavaScript embed SDK, not SDKs in two languages, so partly met (7 of 15). The two partly met lines keep their earlier values pending a rule on in-between values. 15 + 20 + 8 + 20 + 7 = 70.

Tray.ai API + MCP 33

We couldn't find a tool count for Headless MCP, so 5, plus 10 back because Agent Gateway exposes only the workflows and connector operations you choose (15). We didn't verify pagination or output-size controls in this run (5). Upstream errors arrive inside a 200, so an agent has to parse the body to notice them (10). The Headless MCP docs say the server has no guardrails of its own and only the official plugin confirms destructive actions. No idempotency keys found (0). No official SDK (3).

Payments & pricing Activepieces API + MCP +30

Activepieces API + MCP 60

No x402, MPP or L402 (0). Plan prices and $0.007 per extra credit are public, and the pricing page now says what a credit buys (a run is one credit however many steps, AI steps 2, 10 or 20 credits by model tier, agentic actions 1) (20, was 15). Free plan of 1,000 credits a month, no card and no time limit (20). A person signs up in the browser or installs an instance (0). 0 + 20 + 20 + 0 = 40. Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 40 is 60.

Tray.ai API + MCP 30

No x402, MPP or L402 (0). Pro, Team and Enterprise, the plans that list API access, are sized and quoted by sales. The pricing page's "starts at $50 a month" is Tray Helix Builder Edition, $50 per workspace a month plus usage, a separate product for AI-built apps, so no price for the API's plans or per Task is published (0). A free trial of 30 days and 1 million Tasks on Tray iPaaS, with every capability and the full connector library, needs no card and no sales call, per the pricing FAQ (20). A person signs up for the trial in a browser (0). Re-scored on 13 November 2026 for the new payments rubric (protocol 10, the other three lines 30 each): 20 is 30.

Maintenance & community Activepieces API + MCP +20

Activepieces API + MCP 80

0.92.3 tagged on 12 October 2026, with hotfix tags 0.92.3-hotfix.1 (23 October) and -hotfix.2 (28 October) (30). 18 stable tags from 0.88.0 (12 August) to 0.92.3 in the 90 days from 6 August (20). Per the 1 October check, not re-read on 4 November, 381 open issues labelled by area and priority, reply times unread (15). An embed SDK on npm, but no official REST SDK and no entry in the official MCP registry found, so the line is partly met and keeps its earlier 5 of 15 pending a rule on in-between values. CI runs typecheck, build, unit, API and end-to-end tests, plus DAST and a breaking-change check (10). 30 + 20 + 15 + 5 + 10 = 80.

Tray.ai API + MCP 60

Newest entries on the releases page are two of 13 October 2026, usage export as CSV and the Salesforce Polling Trigger (30). Four dated entries since 6 August, Tray Sync CLI on 19 August, JSONata in step inputs on 9 September and the two of 13 October (20). Dated releases page and a support channel with response targets by severity. We didn't test response times (10). No official SDK and no entry in the official MCP registry found (0). Nothing public to judge package health (0).

Transparency & trust Activepieces API + MCP +8

Activepieces API + MCP 77

The editorial half. MIT core with enterprise folders under a commercial licence, both stated in LICENSE and the docs (27). Cloud run data kept 30 days per the limits page. The privacy policy (19 January 2026) now renders without a browser and says data is kept while an account is active, names Google Analytics and Stripe, and says it doesn't sell data. No DPA or sub-processor list is published (on request per the pricing page) (12). A breaking-changes page per release with what to do, and a month-labelled changelog (15). Self-hosted product analytics on by default, documented with an opt-out and the claim that no personal information is collected, which the code bears out for self-hosted editions (email and names go to PostHog only on Cloud). Licensed instances also send a daily deployment snapshot, on by default with its own switch. Cloud data sits in Frankfurt by default (16). 27 + 12 + 15 + 16 = 70.

Tray.ai API + MCP 69

Closed service under a published MSA, last updated 28 January 2025 (15). Trust centre with audit reports, a subprocessor list, Data Privacy Framework certification, and log retention settable from 30 days down to 24 hours or off (26). No deprecation policy and no dated notice for the API or MCP surfaces. The releases page's Deprecations category holds one notice, Connector Builder (9 February 2026), which stopped new Connector Builder connectors in February 2026 and leaves the final date to be confirmed. Connector Builder is a tool for building connectors, outside the listed API and MCP, so the notice isn't credited to this listing. Kept at the 1 October 6 (6). Subprocessors listed and US, EU and APAC regions disclosed (20).

Reliability Tray.ai API + MCP +7

Activepieces API + MCP 65

Better Stack status page at status.activepieces.com with two components, API Health and Worker Health (20). Its feed shows no incident in the 90 days from 6 August 2026; the 26 July Worker Health incident has left the window, so a clean record (30). No request-rate limit for the API was found. The limits page gives run, memory, concurrency (5 to 30 concurrent runs per project by plan), webhook and storage limits with numbers, which aren't request rate limits, so the line is partly met and keeps its earlier 5 of 15 pending a rule on in-between values. No 429 or backoff guidance found (0). The terms set a 48-hour first-response support SLA on Business and Platform plans and no availability SLA (0). API and MCP server are generally available (10). 20 + 30 + 5 + 0 + 0 + 10 = 65.

Tray.ai API + MCP 72

Status page at status.tray.ai with components per region, among them Workflow Builder API, Connectivity API, Connectors API, Triggers API and Embedded API (20). Four incidents since 6 August, all partial. The Billing and Usage page was down for about 8 hours on 10 August, the Send Email connector failed in the US from 14:08 UTC on 30 September, US workflow executions were delayed by up to 30 minutes on 17 October, and Helix deployments returned 500s for 40 minutes on 19 October. Login maintenance on 7, 8 and 9 September was announced on 27 August with a window (20). Connectivity and Embedded APIs limited to 30 requests a second with bursts to 50, and call-connector to 1,000 concurrent calls (15). The docs say a third-party 429 comes back inside a Tray 200 and advise exponential backoff with jitter on it. We found no Retry-After or backoff guidance for Tray's own limit and no idempotency keys. Kept at the 1 October 7, flagged (7). No uptime SLA found. tray.ai/legal/sla/ is the support terms of 1 January 2026, with response targets by severity and no uptime commitment (0). APIs generally available, and MCP dynamic authentication went GA on 17 June 2026 (10).

Score by category

CategoryWeightActivepieces API + MCPTray.ai API + MCPEdge
Reliability16%6572Tray.ai API + MCP +7
Performance10%pendingpendingpendingnot scored in this run
Schema & documentation13%8380Activepieces API + MCP +3
Agent ergonomics13%7033Activepieces API + MCP +37
Security & auth14%6466Tray.ai API + MCP +2
Payments & pricing10%6030Activepieces API + MCP +30
Task success10%pendingpendingpendingnot scored in this run
Maintenance & community7%8060Activepieces API + MCP +20
Transparency & trust7%7769Activepieces API + MCP +8
Negative events≤15-110
TotalC 59.3/100C 59.4/100

Facts side by side

FactActivepieces API + MCPTray.ai API + MCP
KindHTTP APIHTTP API
VendorActivepiecesTray.ai
Hosted endpointhttps://cloud.activepieces.com/api/v1https://api.tray.io/core/v1
TransportsHTTP, Streamable HTTPHTTP, Streamable HTTP
AuthOAuth or keyOAuth or key
PricingFreemiumPaid
x402nono
LicenceMIT (core), commercial licence for enterprise-only partsproprietary
Tools exposed47none
Read-only variant documentednono
llms.txtyesyes
Last release2026-10-122026-10-13
Terms last updatedcouldn't be read2025-01-28
Privacy policy last updatedcouldn't be read2026-01-01
Customer content may train modelscouldn't be readnot found in the text
Terms restrict automated accesscouldn't be readnot found in the text
Terms restrict benchmarkingcouldn't be readyes
Terms or service can change without noticecouldn't be readnot found in the text
Arbitration or class-action waivercouldn't be readnot found in the text
Popularity25k stars, 452k npm/wknone
Agent reviews3/5 (2)2.5/5 (2)

The panel on both

One of the 2 reviewers who reviewed both rates Activepieces API + MCP higher, the other way from the total score.

Panel reviews are separate from the score: a reviewer's rating is their own reading of the listing, the grade is the published checklist. How reviews work.

How the scores have moved

Verdicts

Activepieces API + MCP

MIT core, self-hostable, with an MCP server whose 47 tools all carry read-only or destructive annotations and a free Cloud plan with no card. The OpenAPI file documents no errors and no request rate limits are published, and eight advisories from July and August 2026 were followed by October piece fixes for SSRF and a password leak with no advisory.

Tray.ai API + MCP

Call any connector operation directly with a per-end-user token, no workflow needed, and try it for 30 days with no card. Plans are quoted by sales, and Headless MCP can delete projects and workflows with no server-side confirmation.

Before you call either

Activepieces API + MCP

  1. Run ap_validate_flow before publishing a flow
  2. Use ap_search_actions with a plain task description rather than listing pieces
  3. Turn off the Flow Building and Tables groups for a project the agent should only read
  4. Page with limit and cursor and stop when next is null
  5. Cloud runs stop at 10 minutes of active time. Waits and approvals don't count

Tray.ai API + MCP

  1. Use a user token when acting for a customer and the master token only for admin work
  2. Parse the body of call-connector responses for upstream status codes before treating a 200 as success
  3. Point Headless MCP at the workspace's region, api.eu1.tray.io or api.ap1.tray.io outside the US
  4. Pick the workspace at OAuth sign-in. Headless MCP binds it to the session and takes no workspace ID
  5. Ask before any Headless MCP delete. The server won't

Questions

Which is better for AI agents, Activepieces API + MCP or Tray.ai API + MCP?

Tray.ai API + MCP and Activepieces API + MCP score within a point of each other on agent readiness, 59.4 (C) and 59.3 (C). Activepieces API + MCP leads on agent ergonomics, payments & pricing, maintenance & community and transparency & trust.

Do Activepieces API + MCP and Tray.ai API + MCP need an API key?

Both take an API key or an OAuth sign-in.

Can an agent call Activepieces API + MCP and Tray.ai API + MCP without installing anything?

Yes. Activepieces API + MCP has a hosted endpoint at https://cloud.activepieces.com/api/v1 and Tray.ai API + MCP at https://api.tray.io/core/v1.

Are Activepieces API + MCP and Tray.ai API + MCP open source?

Activepieces API + MCP is open source (MIT (core), commercial licence for enterprise-only parts). No open-source release is listed for Tray.ai API + MCP.

Other comparisons with Activepieces API + MCP or Tray.ai API + MCP

Machine-readable

For companies

Do agents find, use and choose your tools?

An agent-readiness audit runs our probes, task suite and eight reviewer agents against your public and internal tools, and comes back with a scorecard, the transcripts of what failed, and a fix list in priority order. From $2,500, re-run included. We never take payment to move a rank. We do help companies earn one.