SuperAPI AI Directory

repurpose-talk

General & other

View repository

Updated 10-16 03:41

About this skill

Before doing anything else:

  1. Check if ~/.devadvokit.md exists.
  2. If it does, read it silently and use it throughout this skill.
  3. If it does not, stop and tell the user: "I need your DevRel context before I can run this skill. Please run /setup-devadvokit first."

What NOT to Do

This skill produces outlines, angles, and frameworks — not full drafts. Specifically:

  • No full blog posts — outlines only. Full drafts are scope creep and end up generic without heavy editing.
  • No video scripts — that's a different skill.
  • No SEO metadata — separate concern, different skill.

The goal is multiple content directions from one talk, not finished pieces. Editing produces the final versions.


Q&A

Ask these questions one at a time. Wait for each answer before asking the next.

  1. What's the talk title and abstract? (Paste or type — this is required. If you only have a title and need help with an abstract, say so and I'll help you draft one.)

  2. Do you have a script, speaker notes, or transcript? Paste it here or provide a file path. This allows me to ground outputs in your actual content rather than infering from the abstract alone. If you only have the abstract, I can still work with that.

  3. Who's the target audience for the repurposed content, and what's their familiarity level? (Beginner / practitioner / expert — this affects depth and terminology)

  4. What tone should the repurposed content have? (Technical and depth-focused? Conversational and approachable? Thought leadership and industry perspective?)

  5. What conference or event was this presented at, and when? (This gives context and provides "as I presented at X" framing for the repurposed content)


Quality Gates (if transcript/notes provided)

If transcript or notes were provided, process them before generating outputs:

  1. Pull 2–3 pull quotes — direct lines that work as standalone social copy. Memorable, quotable, out-of-context friendly.

  2. Flag demo or live content — anything that won't translate to written formats. Note what needs adaptation or removal.

  3. Note claims needing sources — assertions, data points, or claims requiring a source link before publishing.

If only the abstract was provided, skip this section and work from what you have.


Output

Produce all of the following once the Q&A is complete and quality gates (if applicable) have been processed.

1. Blog post outline

H2 structure with per-section notes (not the full draft). Each section has a brief note on what it covers and the key point it makes. Grounded in the actual talk content if transcript/notes were provided.

2. LinkedIn post angles (5 options)

Each with a different hook strategy:

  • Personal story angle
  • Hot take / contrarian angle
  • Data or observation angle
  • Question angle (sparks discussion)
  • How-to angle (practical takeaway)

Each angle should be distinct in tone and approach. Include a brief note on why this angle works for this content.

3. Tweet thread starters (3 options)

Opening tweet only, with a brief note on thread direction. Each starter should lead in a different way — anecdote, provocative statement, practical tip, etc.

4. Newsletter section

150–200 words, designed to slot into an existing newsletter. Includes a hook, key insight from the talk, and a link/call to action. Written in newsletter voice (conversational, direct).

5. Cold pitch paragraph

One paragraph (4–6 sentences) for republishing on a third-party publication. Includes the pitch angle, why it fits that publication's audience, and a brief credibility anchor. Grounded in the conference context ("As presented at X...").


Content Library Check

Scan the content library in ~/.devadvokit.md. If any existing content covers similar ground:

"This overlaps with [title] ([year]) — consider cross-linking or positioning as an update / deeper dive."

If nothing overlaps, omit this section entirely.


Before presenting any output, read ../../shared/ai-antipatterns.md and silently rewrite any flagged patterns. Do not mention this step to the user.