Skill · Sales and Marketing

Event Post Fact-Check

Draft post-conference posts with every speaker, affiliation, session, sponsorship, photo and hashtag checked against the official program, keeping what the program proves separate from what someone remembers.

At a glance

Event Post Fact-Check is an AI skill for sales & marketing teams that you can use in Claude Code, Codex, or type.com.

Updated

What it does

  • Catches the classic event-post mistakes: a sponsor that wasn't, an outdated affiliation, a photo paired with the wrong talk, an "official" hashtag nobody uses.
  • A claims table labels every fact as Program, Program description, Recollection, Verified or Unverified.
  • Delivers a complete draft the team can correct, not a page of placeholders.

Before you start

  • Web access to the event's program or agenda

See an example

Example output from a sample run. Unless the example says otherwise, company names and figures are sample data.

DRAFT — Dana Lee (personal LinkedIn) — Clinic Ops Summit, Denver

Last week I spoke at Clinic Ops Summit in Denver about why rescheduling eats more front-desk time than no-shows.

After my session, @Priya Shah from @Northfield University came up to talk about her research on waitlist fairness. Her talk the next morning set out a way for clinics to offer cancelled slots without favouring whoever answers the phone first.

Thanks to everyone who stayed for questions.

CLAIMS TABLE
| Claim | Status | Source |
| Dana spoke at Clinic Ops Summit, Denver | Program | agenda p.3, Session 4B |
| Priya Shah, Northfield University | Program + verified | agenda p.5; university faculty page |
| Priya approached Dana after her session | Recollection | Dana's message, Nov 10 |
| Talk "set out a way to..." | Program description | session page (pre-event) |
| Fieldnote sponsored the summit | Unverified → removed | logo on a flyer photo only |

Photos: IMG_2041 → Dana's session (title slide visible) · IMG_2043 unpaired (no signage)
Hashtag: unresolved. No official tag on the event's channels; #ClinicOps is in active use (checked 12 recent posts). Dana to choose.
Please confirm (Dana): you attended Priya's talk; the summary matches what she said in the room.

How to use Event Post Fact-Check

  1. type.com: Choose Use Skill to add Event Post Fact-Check to your type.com workspace.
  2. Claude Code: Download the skill and extract it into ~/.claude/skills/event-post-fact-check/ (or .claude/skills/event-post-fact-check/ in a project). Then ask Claude to use Event Post Fact-Check.
  3. Codex: Download the skill and extract it into ~/.agents/skills/event-post-fact-check/ (or .agents/skills/event-post-fact-check/ in a repository). Then ask Codex to use Event Post Fact-Check.

AI agents can read this page as Markdown or fetch the raw SKILL.md.

Frequently asked questions

What is Event Post Fact-Check?

Event Post Fact-Check is an AI skill for sales & marketing teams that you can use in Claude Code, Codex, or type.com.

What does Event Post Fact-Check do?

Draft post-conference posts with every speaker, affiliation, session, sponsorship, photo and hashtag checked against the official program, keeping what the program proves separate from what someone remembers.

What do I need to use Event Post Fact-Check?

Web access to the event's program or agenda.

Can I use Event Post Fact-Check in Claude Code or Codex?

Yes. Event Post Fact-Check is a standard Agent Skills folder with a SKILL.md file. Download it and extract it into ~/.claude/skills/ for Claude Code or ~/.agents/skills/ for Codex.

Who created Event Post Fact-Check?

Barbara Jovanovic from Startup Cookie created Event Post Fact-Check. The type.com team reviews every skill before it is published in the Skills Library.

  • AI List Inclusion OutreachGet named on the "best X" lists that ChatGPT, Claude, Perplexity and Google AI answers already cite for your buyers' questions. Ranks the lists, drafts a three-line request for each, and tracks follow-ups.
  • Transparent Lower ThirdOn-brand animated name tags for testimonial videos, delivered as transparent ProRes 4444 files that drop straight onto a CapCut or Premiere timeline, with brand pulled from your site and a frame-by-frame check.
  • Social Batch Review CalendarReplace PDFs and chat recaps with a Type app your client can actually review: a weekly calendar with each post's first paragraph and full image, copyable captions, downloadable visuals, and the approver on every card.
  • Reference Style Model"We want our LinkedIn to sound like them." Turns a reference company's public posts into a style spec: numbered moves, each proved by a real line, and a pre-publish checklist. It transfers form, never their facts.
  • Founder GhostwriterGhostwrite LinkedIn and X posts in a founder's real voice, grounded only in things they actually said. Every post has a verbatim anchor, passes six gates, and is read by fresh-context reviewers before you see it.
  • Slop EditLine-edit any draft so it stops reading as AI-written. A mechanical lint catches the surface patterns, a sentence-by-sentence pass fixes agentless and vague constructions, and every reviewer correction becomes a rule.

AI for marketing teams: a practical guide

Browse the technical files
---
name: event-post-fact-check
description: Draft and fact-check social posts about a conference, summit, talk or trade show so every name, affiliation, session, sponsorship, photo and hashtag is verified before publishing. Use for post-event LinkedIn posts by a speaker or attendee, company-page event recaps, and speaker shout-outs. Separates what the official program proves from what someone remembers.
---

# Event Post Fact-Check

Event posts go wrong in predictable ways: a sponsor that wasn't, a speaker's old affiliation, a photo paired with the wrong talk, a summary of a talk the author didn't attend, an "official" hashtag nobody uses. This skill drafts the post **and** produces a claims table showing where every fact came from.

## Inputs
- Who is posting (person or company page) and their role at the event (speaker, attendee, exhibitor).
- What they remember (voice note, chat message, transcript). Treat this as **recollection**, not fact.
- Photos, with the order they were sent.
- Event name and approximate date.

## Step 1. Get the official record
Find the event's own program: agenda PDF, schedule page, session pages, speaker bios. Find the full speaker list yourself; don't ask the user for it, and don't report "search was down" as a reason to skip.

From the program, record: event name, city, venue, dates, the poster's own session (title, slot, room), and every other session or person the post will mention.

## Step 2. Build the claims table
Every factual element in the draft gets a row before you write prose:

| Claim | Status | Source |
|---|---|---|
| Speaker X is at University Y | **Program** | agenda p.4 |
| X's talk was about Z | **Program description** (pre-event) | session page |
| X approached the poster after his talk | **Recollection** | poster's message |
| Company was a sponsor | **Unverified → removed** | flyer logo only |

Status values:
- **Program**: stated in official event materials.
- **Program description**: what the talk was *meant* to cover, written before the event. Not proof of what was said in the room.
- **Recollection**: what the poster remembers. Usable in first person, never stated as general fact.
- **Verified elsewhere**: confirmed on the person's or company's own site.
- **Unverified**: not usable until confirmed. Remove it or ask.

## Step 3. Verification rules
1. **Affiliations:** confirm against the program and the person's current profile. Use the legal or display name they use themselves (site footer, LinkedIn).
2. **Talk summaries:** hedge to the published description. "Her approach helps clients" becomes "Her talk set out an approach meant to help clients" if the description only says she argues for it. Cut synthesis the description doesn't support.
3. **Pairing talks:** link two talks only where both descriptions support the connection, and don't claim either speaker referenced the other.
4. **Agenda ≠ room.** Note in the delivery: "Talk summaries come from the published descriptions, not notes from the room." Ask the poster to confirm they attended each talk mentioned, and which moments actually happened.
5. **Sponsorship, awards and partnerships:** don't state them without direct evidence (sponsor page, organizer post you can open). A search-result preview or a logo on a flyer isn't enough. Remove the claim and keep what you can support ("we had a booth").
6. **Photos:** pair each photo with a talk only from the poster's explicit clarification or visible evidence (signage, a title slide). Use the order the photos were sent as a hint, not proof. A distant shot with no signage stays unpaired. If a post has no usable photo, keep it text-only and say what photo would work (title slide, photo with the speaker, program screenshot).
7. **Tagging:** verify the exact profile before tagging. Tag the person first, then the institution.
8. **Hashtags:** look for an official hashtag on the event's own channels. If there isn't one, use one the organizer actually uses, or a topic tag you've checked is in real use. Never call a hashtag "official" or "established" without evidence; if unresolved, say so and leave the choice to the poster.
9. **Citations:** if the post cites research, check every DOI (for example against Crossref) and flag anything written from memory. Re-check any statistic against the newest edition of its report.
10. **Off-limits:** respect anything the poster has asked to hold back (topics, unannounced customers, numbers).

## Step 4. Write the post
- Write a **complete, standalone draft** the team can correct. Don't fill it with `[placeholder]` gaps.
- Recent events: "last week" or "last weekend". Exact dates only when the date is the news.
- People are the actors in every sentence. The company page post describes what the company does for its customers; it doesn't pivot to the founder's bio.
- Once the announcement post is out, follow-up posts drop the conference framing and stand on the idea.
- Run the slop-edit skill (if installed) before delivery.

## Step 5. Deliver
```
DRAFT — <account> — <event>
<post text>

Claims table: (as above)
Photos: <file> → <post/talk> (basis: <clarification / signage>) · <file> unpaired (why)
Hashtag: <tag> (basis) | unresolved: <options>
Removed: <claim> — <why>
Please confirm: <list of recollection-based lines and open items, by person>
```