---
name: showup-draft
description: Turn a real idea, build moment, decision, failure, or Content Inbox entry into relatable personal social drafts in Steven's voice. Use for X or LinkedIn drafts, quote-post reactions, rewrites, build-in-public posts, and platform adaptations.
---

# Show Up draft

Turn one grounded moment into copy that sounds like Steven.

Read [Steven's social voice](../../references/voice.md) and [platform fit](../../references/platforms.md). When the source needs verification, classification, reference examples, or more evidence, also read [work sources and safety](../../references/sources.md).

## Workflow

1. Apply the proof-to-post filter. Identify what happened, what shipped or broke, the evidence, why somebody might care, and the reusable pattern. If the source lacks the human point, ask only for the missing context that changes the story. If anyone in AI could have written the material, leave it as a seed.
2. Check every factual claim against the supplied material or linked evidence. Mark inference and uncertainty plainly. For private work, use only the generalized pattern. Never invent emotions, motivations, authority, or outcomes.
3. Pick the shape that lets the source do the most work:
   - quote-post reaction for a source post whose joke or verdict needs only a short response
   - real-moment report for something Steven just experienced
   - shipped proof for a concrete artifact or result
   - opinion for a claim grounded in something Steven saw or built
   - technical explanation or Q&A for a concept that needs a short teaching sequence
   - genuine community question when Steven wants real answers
   - event or workshop announcement when the outcome and logistics justify a CTA
   - screenshot or meme as the payload when the asset carries the meaning
   - spicy mode, only when requested and supported by a real contradiction, cost, failure, or artifact
4. Adapt the shape to each requested platform. Preserve the underlying story, not identical copy.
5. Return one strong draft per requested platform. Give variants only when Steven asks or two approaches represent a real decision.
6. Keep the routing internal unless naming the shape helps Steven review the draft. Cut creator language, drafting analysis, and explanations the source already supplies.

When Steven asks to use saved bookmarks or swipe-file examples, use them to inform a pattern—not to imitate an author or invent support for a claim. Select the explicitly named X bookmark folder before reading it. If he names `Tweets`, do not substitute the general Bookmarks feed. Fetch the live Voice / Swipe File schema before saving new examples or feedback.

Search current discussion only when Steven requests trend research or a time-sensitive claim needs verification. Cite any sources used.

## Learn from the decision

When the draft belongs to a Notion content record and Steven approves, edits, or rejects it, fetch the live schema before writing feedback. Preserve the original draft, approved version, what changed, rejection reason, and platform in the properties or page content that exist. Do not create a local log or add guessed properties.

Move a record to `Approved` only when `Human context ready?` and `Safe to share?` are both explicit. Approval applies to the exact copy and platform recorded in Notion.

## Boundaries

- Drafting returns copy in chat. Save it to Notion or Typefully only when Steven requests that write.
- Plane and client work follow the private-material rules in the source reference.

## Done

The draft is complete when one clear human point survives, the source supports every claim, the copy fits the platform and selected shape, and the response contains no unnecessary routing analysis.