Tray.ai platform help — enterprise iPaaS with 700+ connectors, Intelligent iPaaS, Enterprise Core governance, Merlin Agent Builder for AI agents, Tray Embedded for SaaS vendors, GraphQL API, Connector Development Kit. Use when Tray bill keeps climbing and task consumption is unpredictable, workflows fail with unclear errors and debugging feels opaque, evaluating Tray vs Workato vs MuleSoft vs Boomi, embedding integrations into a SaaS product via Tray Embedded, building Merlin AI agents, or configuring the GraphQL Embedded API and solution instances. Do NOT use for simple Zapier/Make automations (use /sales-integration), Workato-specific questions (use /sales-workato), or MuleSoft-specific questions (use /sales-mulesoft).
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
A direct command skips the review prompt. Inspect the source before running it.
The command stays on one line. Scroll horizontally to inspect it before copying.
Prefer a local copy? Download the files currently available to SkillsMP.
File Explorer
4 files
Showing SKILL.md
SKILL.md
Source instructions · Read-only preview
name
sales-tray
description
Tray.ai platform help — enterprise iPaaS with 700+ connectors, Intelligent iPaaS, Enterprise Core governance, Merlin Agent Builder for AI agents, Tray Embedded for SaaS vendors, GraphQL API, Connector Development Kit. Use when Tray bill keeps climbing and task consumption is unpredictable, workflows fail with unclear errors and debugging feels opaque, evaluating Tray vs Workato vs MuleSoft vs Boomi, embedding integrations into a SaaS product via Tray Embedded, building Merlin AI agents, or configuring the GraphQL Embedded API and solution instances. Do NOT use for simple Zapier/Make automations (use /sales-integration), Workato-specific questions (use /sales-workato), or MuleSoft-specific questions (use /sales-mulesoft).
argument-hint
[describe what you need help with in Tray]
license
MIT
version
1.0.0
tags
["sales","integration","ipaas","api","platform"]
github
https://github.com/trayio
Tray.ai Platform Help
Step 1 — Gather context
If references/learnings.md exists, read it first for accumulated platform knowledge.
Ask the user:
What do you need help with?
A) Workflow building — connectors, triggers, steps, data mapping
B) Merlin Agent Builder — AI agents, knowledge, tools, MCP
C) Tray Embedded — embedding integrations in your SaaS product (Solutions, Solution Instances, Configuration Wizard)
D) API Management — rate limiting, throttling, governance
E) Connector Development Kit (CDK) — building custom connectors
F) GraphQL API — Embedded API, Master/User tokens, mutations
C) Admin / architect (governance, security, deployment)
D) AI/automation builder (Merlin agents)
E) Evaluating Tray
Skip-ahead rule: if the user's prompt already provides enough context, skip to Step 2. Lead with your best-effort answer using reasonable assumptions (stated explicitly), then ask 1-2 critical clarifying questions.
Step 2 — Route or answer directly
Problem domain
Route to
Connecting Tray to Salesforce/HubSpot/CRM
/sales-integration {question}
CRM comparison — Salesforce vs HubSpot vs others
/sales-crm-selection {question}
Salesforce-specific config (Flows, Apex, SOQL)
/sales-salesforce {question}
Simple no-code Zapier/Make automations
/sales-integration {question}
Workato-specific questions
/sales-workato {question}
MuleSoft-specific questions
/sales-mulesoft {question}
Boomi-specific questions
/sales-boomi {question}
SnapLogic-specific questions
/sales-snaplogic {question}
Celigo-specific questions
/sales-celigo {question}
Jitterbit-specific questions
/sales-jitterbit {question}
CRM data quality / deduplication
/sales-data-hygiene {question}
Workflow design and task optimization
Answer directly — read platform guide
Tray Embedded integration
Answer directly — read platform guide
Merlin Agent Builder
Answer directly — read platform guide
GraphQL API / solution instances
Answer directly — read API reference
Pricing and cost optimization
Answer directly — read platform guide
When routing, provide the exact command.
Step 3 — Tray platform reference
Read references/platform-guide.md for the full platform reference — modules, pricing, connectors, Merlin agents, Tray Embedded, rate limits, governance.
Answer the user's question using only the relevant section. Don't dump the full reference.
If the question involves the GraphQL Embedded API (solution instances, user tokens, authentications), also read references/tray-api-reference.md for endpoints, authentication, and mutation details.
Step 4 — Actionable guidance
You no longer need the platform guide — focus on the user's specific situation.
Connector selection shortcut: Check Tray's connector library first (700+ pre-built). If no connector exists, use the HTTP Client connector for any REST API, or build a custom connector with the Connector Development Kit (falafel Node.js framework).
Cost optimization shortcut: Task-based billing — each workflow step counts. Audit high-volume workflows, remove redundant steps, use branching to skip unnecessary actions. Enterprise contracts typically start at $36K/year; model task consumption before signing and negotiate usage caps.
Debugging shortcut: Use the workflow Debug panel to step through execution and inspect input/output at each step. For the Embedded API, inspect GraphQL responses for errors array — Tray returns HTTP 200 with an errors field even on failure. Check connection auth status; expired auths silently fail.
Tray Embedded shortcut: Solution instances are per-end-user copies of a Solution. Use the Master Token for org-wide setup (manage end users, create user tokens via the authorize mutation, fetch connectors/schemas). Use a User Token for per-end-user operations — including the entire Solution Instance lifecycle (createSolutionInstance, updateSolutionInstance to enable/rename, upgradeSolutionInstance for config/auth values, removeSolutionInstance). Get Solution Instances accepts either token. All API calls must be server-side — CORS blocks browser calls.
If you discover a gotcha, workaround, or tip not covered in references/learnings.md, append it there.
Gotchas
Best-effort from research — review these, especially pricing details and feature availability which change frequently.
No public list prices. Tray is sales-led — no self-service signup. The official pricing page (re-verified 2026-06-13) lists three editions — Pro / Team / Enterprise — but publishes no prices; cost = base plan + usage (Tasks) + add-ons, with quarterly overage billing. Pro = 3 workspaces / 7-day insights; Team = 20 workspaces / 30-day insights / add-ons purchasable; Enterprise = unlimited workspaces / 180-day insights / SSO + HIPAA + regional hosting + log streaming. Third-party reports cite ~$595/mo (25K tasks) Pro and ~$1,800/mo (100K tasks) Team — directional only, not confirmed by Tray. Enterprise commonly starts ~$36K/year. Always get a written quote.
Task-based billing escalates unpredictably in production. The most common user complaint across G2/Capterra reviews: task consumption in production exceeds estimates. Model expected task volume carefully and ask Tray sales for consumption alerts and soft caps before signing.
GraphQL backend-only — CORS blocks client-side calls. "The Tray GraphQL API is a backend API only and calls must be invoked server-side." You cannot call the Embedded API from a browser. All integration UI in your SaaS product must route through your backend.
Two token types with different scopes. Master Token = org-wide setup (Get Users, create/update/delete external users, mint user tokens via authorize, fetch connectors/schemas). User Token = single end user — and the entire Solution Instance lifecycle runs on the User Token (createSolutionInstance, updateSolutionInstance, upgradeSolutionInstance, removeSolutionInstance are all "(user token)" in current docs). Get Solution Instances accepts either token. Use the wrong token and you'll get empty results or auth errors.
Rate limits: 30 req/sec, 1800 req/min, burst 50/sec. Call connector endpoint has separate concurrency limit of 1000 active requests. Docs don't specify which HTTP response code is returned when exceeded — monitor for rate-limit errors and implement exponential backoff.
Documentation gaps are a consistent complaint. G2 and Capterra reviewers repeatedly note thin docs, missing real-world examples, and steep learning curve for advanced features. Budget training time and pair new builders with experienced ones.
Regional base URLs differ. US: tray.io/graphql. EU: eu1.tray.io/graphql. APAC: ap1.tray.io/graphql. Whitelabel OAuth redirects use {org}.integration-authentication.com/oauth2/token. Using the wrong region returns empty results without an obvious error.
Tray.io → Tray.ai rebrand. The product rebranded to Tray.ai in 2024 with heavy AI/Merlin pivot. tray.io now 301-redirects to tray.ai, but API surface and developer docs still use tray.io domains. Both work in documentation links.
/sales-salesforce — Salesforce platform help — common Tray integration target
/sales-hubspot — HubSpot platform help — common Tray integration target
/sales-data-hygiene — CRM data quality — deduplication, record matching, enrichment automation
/sales-do — Not sure which skill to use? The router matches any sales objective to the right skill. Install: npx skills add sales-skills/sales --skill sales-do
Examples
Example 1: Tray bill climbing and we can't see where the tasks are going
User says: "Our Tray bill went from $40K to $65K in 9 months — how do we figure out what's consuming tasks?"
Skill does: Reads platform guide for pricing section. Explains task-based billing — every workflow step counts, so high-volume workflows with many branches/steps are usually the culprits. Recommends: (1) review the workflow task consumption view in the Tray dashboard, (2) audit the top 10 highest-volume workflows for redundant or always-on steps, (3) add branching to skip unnecessary actions, (4) batch-process where the source API supports it, (5) renegotiate at contract renewal with usage data in hand. Notes that Tray doesn't hard-stop workflows at plan limits, so overages compound.
Result: User has a concrete optimization plan and a starting audit list.
Example 2: Embedding Tray Solutions in a B2B SaaS product
User says: "We're a SaaS company and want to embed Tray so our customers can connect their own tools. How does that work?"
Skill does: Reads platform guide for Tray Embedded section and API reference for mutation details. Explains: (1) build a Solution in Tray (reusable workflow template), (2) use the Configuration Wizard to let end users authenticate and configure, (3) on your backend, first use the Master Token to register the end user (createExternalUser) and mint a per-user token via the authorize mutation, (4) with that User Token, call createSolutionInstance (solutionId, instanceName) to create the per-customer copy, then updateSolutionInstance with enabled: true to activate it (config/auth values are applied via upgradeSolutionInstance), (5) note the ~2-second delays required between create → enable → first webhook call. Flags CORS constraint (API is backend-only), and points to the official sample app (github.com/trayio/embedded-edition-sample-app) for reference architecture.
Result: User understands the embedded integration model, token types, and reference implementation.
Example 3: Evaluating Tray vs Workato for a technical team
User says: "We're a 6-person platform engineering team integrating 30 SaaS apps. Tray or Workato?"
Skill does: Reads platform guide for competitor comparison. For a technical team integrating many apps: Tray fits better for teams comfortable with API-first, JavaScript-adjacent workflows and custom connector development via the CDK. Workato fits better for teams that want faster time-to-value with business-user-friendly recipes and broader connector coverage (1,200+ vs 700+). Pricing is similar enterprise range. Notes: Tray docs are thinner — budget more ramp time. If you plan to embed in a product, Tray Embedded is mature (launched 2020); Workato also has Embedded iPaaS with the Embed Partner Program.
Result: User understands trade-offs and can pick based on team preference and roadmap.
Troubleshooting
Workflow silently fails — no error shown, no output
Symptom: A workflow step completes without an error but downstream steps receive empty data.
Cause: Tray's GraphQL API returns HTTP 200 with an errors field when the server encounters a problem. Connection authentication may have expired without flagging, or the step's input mapping references a null/missing field.
Solution: Open the workflow Debug panel and inspect input/output at each step. Check Connections → verify the connection for the step shows "Authenticated" (not expired). Re-authenticate if needed. For API-level debugging, inspect the full GraphQL response — look for errors array even on HTTP 200 responses.
Getting 429 or rate-limit errors at scale
Symptom: Workflow actions fail intermittently when processing high volumes.
Cause: Tray enforces 30 req/sec and 1800 req/min across Embedded/Connectivity APIs (except Call Connector). Bursts up to 50/sec are tolerated briefly. Hitting these limits returns rate-limit responses (specific HTTP code not published).
Solution: Implement exponential backoff on rate-limit errors. For high-volume Call Connector usage, remember concurrency cap is 1000 active requests — request a CSM increase if needed. Reduce trigger frequency or batch source events. For Event delivery (Trigger API), Tray auto-implements exponential backoff when downstream endpoints rate-limit.
Solution Instance not appearing for a specific end user
Symptom: You created a Solution Instance via the Embedded API but the end user doesn't see it in their Configuration Wizard.
Cause: Wrong token scope or ownerId, or querying from the wrong regional endpoint. Get Solution Instances (get-solution-instances) accepts a user or master token and takes an ownerId argument — a master token returns instances org-wide, a user token returns only that end user's. A mismatched ownerId (or a user token for a different end user) returns nothing. Also: if the instance was just created, the required ~2-second create→enable delay may not have elapsed, so it isn't enabled yet.
Solution: Either query with the master token filtered by the correct ownerId, or mint a User Token for that end user via the authorize mutation (pass their userId) and query Get Solution Instances with it. Confirm the instance is enabled (set via updateSolutionInstance enabled: true). Confirm the regional base URL matches your org's region (US: tray.io, EU: eu1.tray.io, APAC: ap1.tray.io). Whitelabel deployments use {org}.integration-authentication.com for OAuth redirects — verify the end user's auth flow hits the correct redirect URL.
