Snap Studio Problems¶

Outcome: Understand Snap Studio's normal "not quite there yet" states — and what to do at each one. Audience: Everyone / Developers · Tags: troubleshooting, studio, ai

Most things that feel like a "failure" in Snap Studio are actually the assistant working correctly and asking you for help — Studio never guesses silently. This page walks through each one, symptom-first.

"I described what I wanted, and got a question back instead of a finished plan"¶

This is normal, not an error. If the assistant couldn't figure out a piece of information from your description — most commonly a destination — it leaves that field blank and surfaces the gap as a question in a highlighted "I need to know:" box at the top of Stage 2 (Review & edit), for example "What email address should the message go to?"

Fix: You don't answer these questions in a separate box. Find the field the question is pointing at on the matching step card (e.g. the "Send it to" field) and fill it in there. Once every question's matching field has a value, click Check my flow.

If you keep getting asked the same kind of question across different prompts, make your description more specific — include the actual destination (an email address, or an http(s):// URL) directly in your sentence rather than describing it abstractly. See Describe it for examples of specific-enough prompts.

"Check my flow found problems"¶

Clicking Check my flow validates your current plan. If it comes back "ok": false, you get a list of problems — plain-English sentences that each name the step and (when relevant) the field to fix:

{
  "ok": false,
  "problems": [
    { "stepId": "action-1", "field": "to", "message": "Then: send an email: please fill in \"Send it to\"." }
  ]
}

Fix: Read each problem — it already tells you exactly what's wrong and where. Fill in or correct the named field, then click Check my flow again. Your keyboard focus jumps straight to the first problem's field automatically, and the field itself shows an inline error right where you're looking, so you shouldn't need to hunt for it. Repeat until you see "Looks good — ready to save."

"My test run failed"¶

Save & test it (Stage 3) does a real thing: it saves your flow, then runs it once, for real through the actual execution engine — if the action is "send an email," an email actually goes out. If that real run fails, you'll see an alert card instead of the green success confirmation:

{
  "status": "failed",
  "headline": "We hit a problem running your flow.",
  "problem": {
    "message": "SnapLink couldn't send the email — the mail server isn't set up yet.",
    "suggestion": "Ask your SnapLink admin to configure the email connector, then try again."
  }
}

Fix: Read problem.message (the plain-language cause) and problem.suggestion (what to do about it) on the result screen — these map to the same failure taxonomy as any other run (e.g. an unset SMTP configuration surfaces as an auth-shaped problem). Click Go back and edit to return to Stage 2 with your plan fully intact, fix the field or ask your admin to fix the underlying setup, then try Save & test it again. Your flow was still saved from this attempt — it just didn't run cleanly yet, so nothing is lost by going back.

This is different from a flow that simply hasn't been triggered yet — a failed test run means the flow did try to run and something in the attempt genuinely broke. If you're instead wondering why a live flow isn't firing at all, see Flow not running.

"The plan I sent got rejected" (developers)¶

If you're calling Snap Studio's API directly rather than through the UI, you may see one of two errors when submitting a plan to /studio/check or /studio/approve:

STUDIO_PLAN_MALFORMED (400) — the plan doesn't even have the right shape: it failed schema validation, or broke a structural rule (must start with exactly one trigger step, followed by one or more action steps). This usually means a plan was hand-edited or generated outside the normal Describe → Review flow.
STUDIO_PLAN_INVALID (400) — the plan is well-formed but didn't pass validation; the response includes the same problems array described above under "Check my flow found problems", also covered in Review & edit.

Both are because client-sent plans are never trusted — every plan is re-validated server-side on every call, regardless of what the client claims. This also means Studio plans are restricted to a fixed, whitelisted set of connector/capability pairs: a webhook trigger, plus rest_sdk (request) or email (send_email) actions. Any other connector or capability in a submitted plan is rejected as malformed — Studio doesn't draft (or accept) plans using other connectors yet.

Fix: Don't hand-construct or mutate plan JSON outside what Stage 2's own edits produce. If you're integrating against the API directly, always POST /studio/check after any edit and use its response — including a fresh problems list — as the source of truth before calling /studio/approve.

Was this helpful? 👍 👎