feat(server-utils): Instrument the OpenAI decisions API as gen_ai.decide spans

[andreiborza]

What

OpenAI decisions.create calls now create decide <model> spans with op gen_ai.decide, recorded in the same shape as Jev calls so Sentry renders them the same way:

• input is recorded as state, images included as they are.
• Questions and answers are keyed by question name, as the conventions define.
• Choice and score question options go in criteria, like Jev: choices as { [value]: description }, score levels as their labels.
• Choice and score probabilities are keyed by option or score value, and score labels go in a legend.
• predicate answers keep OpenAI's { type, probability }, which Sentry renders as yes/no.

	OpenAI sends / returns	Recorded on the span
Evaluated input	input: "I want a refund"	state: "I want a refund"
Questions	[{ name: "wants_refund", type: "predicate", ... }]	{ wants_refund: { type: "predicate", ... } }
Choice question	choices: [{ value: "billing", description: "Charges and refunds" }, { value: "other" }]	criteria: { billing: "Charges and refunds", other: "" }
Score question	levels: [{ label: "low" }, { label: "high", description: "Blocks the customer" }]	criteria: ["low", "high"], descriptions: { "1": "Blocks the customer" }
Answers	[{ name: "wants_refund", type: "predicate", ... }]	{ wants_refund: { type: "predicate", ... } }
Predicate answer	{ type: "predicate", probability: 0.97 }	unchanged
Choice answer	choice: true, probabilities: [{ value: true, probability: 0.91 }, ...]	choice: "true", probabilities: { "true": 0.91, ... }
Score answer	probabilities: [{ value: 1, label: "high", probability: 0.95 }, ...]	probabilities: { "1": 0.95, ... }, legend: { "1": "high", ... }

OpenAI decisions have a few things Jev does not:

• predicate questions and answers instead of noul, with the yes probability in probability.
• Choice values can be booleans as well as strings. The answer's choice is recorded as a string so Sentry renders it.
• Score levels can have descriptions, recorded in descriptions keyed by score value. Sentry does not render them yet.

  "urgency": {
    "type": "score",
    "instructions": "How urgent is this ticket?",
    "criteria": ["low", "medium", "high"],
    "descriptions": { "1": "Needs a reply today", "2": "Blocks the customer" }  <-- Jev does not have these
  }

• A question can come back as a refusal, { type: "refusal" }, with no probability or reason. Sentry does not render it yet.

  "answers": {
    "wants_refund": { "type": "predicate", "probability": 0.97 },
    "urgency": { "type": "refusal" }  <-- Jev does not have these 
  }

• The input can hold images as base64 data URLs.

  "state": [
    {
      "role": "user",
      "content": [
        { "type": "input_text", "text": "Is this image mostly red?" },
        { "type": "input_image", "image_url": "data:image/png;base64,iVBORw0KGgo..." }  <-- Jev does not have these
      ]
    }
  ]

Why

OpenAI added a Decisions API that works like Jev, and these calls had no span.

Related: #25253

[cursor]

Stale Bugbot comment from a previous run.

[github-actions]

size-limit report 📦

⚠️ Warning: Base artifact is not the latest one, because the latest workflow run is not done yet. This may lead to incorrect results. Try to re-run all tests to get up to date results.

Path	Size	% Change	Change
@sentry/browser	29.83 kB	-	-
@sentry/browser - with treeshaking flags	27.96 kB	-	-
@sentry/browser - with treeshaking flags tracing without tracing	27.87 kB	-	-
@sentry/browser (incl. Tracing)	52.02 kB	-	-
@sentry/browser (incl. Tracing + Span Streaming)	52.02 kB	-	-
@sentry/browser (incl. Tracing, Profiling)	54.96 kB	-	-
@sentry/browser (incl. Tracing, Replay)	91.75 kB	-	-
@sentry/browser (incl. Tracing, Replay) - with treeshaking flags	80.62 kB	-	-
@sentry/browser (incl. Tracing, Replay with Canvas)	96.47 kB	-	-
@sentry/browser (incl. Tracing, Replay, Feedback)	109.43 kB	-	-
@sentry/browser (incl. Feedback)	47.35 kB	-	-
@sentry/browser (incl. sendFeedback)	34.89 kB	-	-
@sentry/browser (incl. FeedbackAsync)	40 kB	-	-
@sentry/browser (incl. Metrics)	30.85 kB	-	-
@sentry/browser (incl. Logs)	31.13 kB	-	-
@sentry/browser (incl. Metrics & Logs)	31.78 kB	-	-
@sentry/react	31.64 kB	-	-
@sentry/react (incl. Tracing)	54.34 kB	-	-
@sentry/vue	37.98 kB	-	-
@sentry/vue (incl. Tracing)	54.95 kB	-	-
@sentry/svelte	29.86 kB	-	-
@sentry/remix (Remix 3 client bundle)	57.02 kB	-	-
CDN Bundle	31.56 kB	-	-
CDN Bundle (incl. Tracing)	52.52 kB	-	-
CDN Bundle (incl. Logs, Metrics)	33.77 kB	-	-
CDN Bundle (incl. Tracing, Logs, Metrics)	54.46 kB	-	-
CDN Bundle (incl. Replay, Logs, Metrics)	74.7 kB	-	-
CDN Bundle (incl. Tracing, Replay)	90.19 kB	-	-
CDN Bundle (incl. Tracing, Replay, Logs, Metrics)	92.11 kB	-	-
CDN Bundle (incl. Tracing, Replay, Feedback)	96.34 kB	-	-
CDN Bundle (incl. Tracing, Replay, Feedback, Logs, Metrics)	98.28 kB	-	-
CDN Bundle - uncompressed	92.98 kB	-	-
CDN Bundle (incl. Tracing) - uncompressed	155.83 kB	-	-
CDN Bundle (incl. Logs, Metrics) - uncompressed	99.52 kB	-	-
CDN Bundle (incl. Tracing, Logs, Metrics) - uncompressed	161.79 kB	-	-
CDN Bundle (incl. Replay, Logs, Metrics) - uncompressed	229.56 kB	-	-
CDN Bundle (incl. Tracing, Replay) - uncompressed	276 kB	-	-
CDN Bundle (incl. Tracing, Replay, Logs, Metrics) - uncompressed	281.94 kB	-	-
CDN Bundle (incl. Tracing, Replay, Feedback) - uncompressed	289.7 kB	-	-
CDN Bundle (incl. Tracing, Replay, Feedback, Logs, Metrics) - uncompressed	295.63 kB	-	-
@sentry/nextjs (client)	56.75 kB	-	-
@sentry/sveltekit (client)	52.41 kB	-	-
@sentry/core/server	41.12 kB	-	-
@sentry/core/browser	13.81 kB	-	-
@sentry/node	152.43 kB	+0.31%	+460 B 🔺
@sentry/node/import (ESM hook with diagnostics-channel injection)	83.81 kB	+0.04%	+30 B 🔺
@sentry/node - without tracing	94.57 kB	+0.04%	+34 B 🔺
@sentry/node - without channel injection	130.66 kB	+0.38%	+484 B 🔺
@sentry/aws-serverless	102.73 kB	+0.04%	+39 B 🔺
@sentry/cloudflare (withSentry) - minified	210.38 kB	-	-
@sentry/cloudflare (withSentry)	521.61 kB	-	-
@sentry/nextjs/cloudflare (withSentry) - minified	228.06 kB	-	-

View base workflow run

[cursor]

Stale Bugbot comment from a previous run.

[cursor]

Cursor Bugbot has reviewed your changes and found 2 potential issues.

^{❌ Bugbot Autofix is OFF. To automatically fix reported issues with cloud agents, enable autofix in the Cursor dashboard.}

^{Reviewed by Cursor Bugbot for commit 0cf6661. Configure here.}

[nicohrubec]

thanks for updating, lgtm

[JPeer264]

{ type: 'predicate', name: 'lgtm', probability: 0.99 }