How should you approve one delivered-account video slot after a dry run?

This article addresses a narrower problem than a general dry-run explainer: how to review one existing delivered-account video slot after simulation, with a reusable approval artifact. TokPortal documents that dry run performs the same validation and server-side pricing as a real write, then stops before the first write. Nothing is created, nothing is charged, and any identifiers returned are synthetic planning artifacts rather than live objects [1]. That contract is useful, but teams still need an internal decision method for the specific slot they intend to spend.

The delivered-account MP4 preview recipe describes an existing-account operation. The documented recipe starts only after a real account has already been delivered and already has a video-capable bundle with an unused slot. The check does not purchase capacity, store media, or publish. Instead, it simulates one configuration request against a real bundle position while keeping the slot unchanged [2]. That makes the relevant review unit very concrete: one account, one bundle slot, one media URL, one description, one disclosure choice, and one target date.

That focus matters because dry run validates structure, not editorial merit. The API can tell you whether the request shape, permissions, and date rules are acceptable for this operation. It cannot assess whether the hook is clear, whether the on-screen proof is persuasive, whether the footage is boring, or whether the post fits the account strategically. Those remain human decisions. A useful approval workflow therefore separates two layers: platform-checked validity and human-judged suitability.

TokPortal’s recipe also warns against incomplete previews. The documented request uses a PUT call for a specific bundle video position and includes video_type, video_url, description, target_publish_date, ai_content_disclaimer, and auto_publish: false [2]. Because this operation is documented as full-replace configuration, the preview should include the fields you actually intend to send later, not a minimal stub that passes validation but omits the real intended state [2]. In practice, that means the safest preview is a rehearsal of the full slot payload, not a loose test of one caption string.

Scheduling is the main place where operators misread what a failure means. TokPortal’s MCP documentation states that one bundle maps to one account, that the earliest target_publish_date is today plus three days while account creation is still pending and today plus one day once the account is delivered or already existing, and that a maximum of three videos per day applies per bundle [3]. The delivered-account preview page further says the API may return INVALID_DATE with details.earliest_allowed, and that response should be treated as TokPortal’s validation result for this request, not as a judgment about the creative itself [2][3]. If the preview fails on date rules, the next move may be rerouting the post to another day, not rewriting the video concept.

Identifier hygiene is the second frequent failure. TokPortal documents that dry-run IDs are synthetic, recognizable, and accepted only inside further dry-run chains; a live request using one is rejected with DRY_RUN_ID_IN_LIVE_REQUEST [1][3]. So the review record should never mix preview references with live execution IDs. If a person approves the post after preview, the later real call must stand on its own returned live identifiers. This is less glamorous than creative review, but it prevents a common handoff error: treating a simulated object as if it already exists in production.

A small approval artifact solves most of this confusion. Instead of asking, 'Did the dry run work?', ask five narrower questions: Did we target the correct existing account? Is the slot genuinely unused and video-capable? Does the preview payload match the intended live payload? Did TokPortal validate the date and field rules for this operation? Has a human explicitly approved live execution as a separate step? That set is specific enough to use immediately and narrow enough to avoid turning dry run into a vague feeling of safety.

Completed worked artifact — filled decision rows for one proposed preview:

Proposed data

• Account state: delivered existing account
• Bundle slot: position 2 on a real video-capable bundle
• Media URL: https://cdn.example.com/videos/flashcards-demo-v3.mp4
• Description: I kept mixing up these four terms, so I turned them into one mnemonic.
• AI-content disclaimer: true
• Target publish date: 2026-09-12
• Auto publish: false

Human decisions

• Rights checked for the MP4: Yes
• Account fit reviewed by operator: Yes
• Live publish approved now: No, preview only

Decision table

1. Destination account correct? — Yes — Reason: existing delivered account chosen intentionally for this audience.
2. Unused video slot confirmed? — Yes — Reason: operator verified position 2 is unused before preview [2].
3. Full payload included? — Yes — Reason: request includes video type, URL, description, date, disclosure, and auto_publish: false [2].
4. Date rule satisfied? — Proposed Yes — Reason: example date is more than one day ahead for a delivered or existing account, but the actual API response remains the deciding check [2][3].
5. Creative approved? — Undecided — Reason: API validation does not answer creative quality.
6. Safe for live execution immediately? — No — Reason: preview success would justify a separate human go/no-go step, not automatic publication.
7. Preview IDs reusable live? — No — Reason: synthetic dry-run IDs must not enter a real call [1][3].

Compact request example — reusable REST payload for the preview itself:

PUT /api/ext/bundles/11111111-2222-4333-8444-555555555555/videos/2`
Headers:
- `X-API-Key: <redacted>`
- `X-TokPortal-Dry-Run: true`
- `Content-Type: application/json

Body: { "video_type": "video", "video_url": "https://cdn.example.com/videos/flashcards-demo-v3.mp4", "description": "I kept mixing up these four terms, so I turned them into one mnemonic.", "target_publish_date": "2026-09-12", "ai_content_disclaimer": true, "auto_publish": false }

What this example is for: structural validation of one proposed slot configuration. What it is not: proof of media durability, proof of publication, proof of audience response, or proof that the creative is strong.

A practical reading rule follows from the docs. Success means TokPortal accepted the request shape for this operation under current validation rules, with dry-run markers present and credits_charged at zero if that field appears [1][2]. Failure means the live call would fail for the same reason if sent unchanged [1]. Neither outcome settles the marketing question. A pass says 'this slot configuration is structurally admissible now.' A human still has to decide whether spending the slot is wise.