ProPost
Empezar
Todos los skills

Marca personal

Thought leadership posts

propost-thought-leadership-posts

Cuándo ayuda

Crafts high-performing LinkedIn-style thought leadership posts in ProPost (sharp hook, clear point of view, proof, tight structure, a real ending) by writing a precise idea brief for generate_draft and then fixing weak spots with rewrite_post or update_post. Use for "write a thought leadership post", "turn my opinion into a LinkedIn post", "make this post better", "my posts get no engagement", "write about X in my voice", "draft a post about what I learned", or any request for an expert-sounding single post.

Los skills funcionan con el conector de ProPost, disponible en Pro y Premium. Configurar el conector

Produces one post (and its per-network cuts) that says something specific, sounds like the person, and survives an editor's read. The leverage is in the brief you give the writer and in the targeted fixes after.

Before you start

  1. get_account: drafts left, autonomy, time zone.
  2. list_channels: max_length per network (LinkedIn is long, X short).
  3. list_personas: which voice each channel uses. If the voice is empty or generic, offer propost-brand-voice first.
  4. list_posts with status: "published" (limit 20): what they posted lately, to avoid repeating, and which posts have the best engagement.
  5. list_ideas: pick from the bank if they have not named a topic.

Workflow

  1. Find the point of view. A post needs a claim someone could disagree with. Ask: "What do you believe about this that most people in your field would push back on?" and "What happened that made you believe it?" If they have no claim, pick a framework from references/frameworks.md that fits their material.
  2. Collect proof the person actually has: a number, a before and after, a specific scene, a client situation they can share. Mark gaps as [add your number]; never fill them.
  3. Write the brief (this is the idea, 10 to 2000 characters, aim for 400 to 800). Include: the claim, the story or proof, the audience, the framework, the ending you want, what to avoid. See the template below.
  4. Generate: generate_draft with idea, channel_ids (all networks the post goes to; they share one draft and are condensed for shorter ones), and persona_id only if overriding the channel's voice. It uses 1 AI draft; say so first. It takes 10 to 40 seconds.
  5. Edit as an editor, using the checklist below. Fix with the cheapest tool: update_post for a line you can fix yourself, rewrite_post (max 3 per post, give one instruction at a time) for structural changes such as "rewrite the first two lines to start with the 40% figure and cut the greeting".
  6. Tailor networks: for X, Threads or Bluesky, update_post on the copy with the single sharpest line (or the first three paragraphs), within max_length. Don't paste LinkedIn text.
  7. Check: check_post; fix every issue.
  8. Deliver by autonomy: drafts level, say it waits in ProPost; schedule level, add_to_queue or schedule_post once they have approved the final text; publish_now only on an explicit yes to the shown text and channel.

Brief template

Claim: Most onboarding emails fail because they teach the product, not the first win. Story: [the real story: what we changed, in what week, what happened to activation, with the real number]. Audience: product managers at early-stage SaaS. Framework: contrarian take with proof. Ending: one practical step they can do this week. Avoid: "game-changer", rhetorical questions as the opener, a list of 10 tips.

A brief with a claim, a story, an audience and an ending gets a specific post. "Write about onboarding" gets a pamphlet.

Writing guidance

Hook (first two lines; LinkedIn shows about 210 characters before "see more"). Make a promise the rest keeps. Patterns that work: a result ("We cut churn from 6% to 3.5% by removing a feature."), a confession ("I fired our best salesperson. He was right to leave."), a contradiction ("The best hire I made had no experience."), a scene ("Tuesday, 9:12. The client called to cancel."). Weak: "I'm excited to share", a question with an obvious answer, a quote from someone famous.

Structure. Hook, then one-sentence stakes, then the story or evidence in short paragraphs (1 to 3 lines, blank line between), then the lesson stated once, then the ending. Cut any paragraph that repeats the hook.

Proof beats assertion. "Retention matters" is a slogan. "In March, 31% of trial users never opened a second session. We emailed them one thing." is evidence. One concrete detail per post is the minimum.

Point of view. State a position and its cost: "This only works under 20 people." Qualifying a claim once makes it more credible; qualifying every sentence makes it nothing.

Ending. Pick one: an instruction ("Try it on your next 1:1."), a sharper restatement of the claim, or a real question they want answers to ("Which meeting would you cut first?"). Not "Thoughts?" every time, not a hashtag wall (0 to 3 tags, via set_hashtags if wanted).

Length. 900 to 1,500 characters is the sweet spot for most LinkedIn text posts; go longer only when each paragraph earns its place.

Editor's checklist before `check_post`:

  • First line stands alone and is specific.
  • One idea. No "and also".
  • No em dashes, no filler openers, no "leverage", "unlock", "delve".
  • Any number or claim traces to something the person told you.
  • It sounds like their samples, not like a brochure.

Frameworks with one-line examples: references/frameworks.md.

Guardrails

  • Never invent results, clients, quotes, statistics or credentials. Use [placeholders] and list them for the person.
  • No private data about others. Don't name clients or colleagues unless the person says it is public and fine.
  • Don't write attacks on named people or companies.
  • Don't regenerate repeatedly; one generate_draft, then targeted edits.
  • Don't publish_now without an explicit yes to the exact text.

If something fails

  • quota_exceeded: write the post yourself from the brief and create_draft.
  • too_long: rewrite_post "shorter, under N characters" or edit; check_post.
  • generation_failed: retry once with a tighter brief; otherwise create_draft.
  • Rewrite limit reached (3 per post): edit with update_post.
  • channel_needs_reconnect: ask them to reconnect; skip that channel.
  • Anything else: see ../CONVENTIONS.md.

Archivos de referencia

Post frameworksreferences/frameworks.md

Pick the one that fits the material the person actually has. Each shows the shape and a one-line example (examples are illustrative; use the person's real facts only).

1. Contrarian take with proof

Shape: common belief, why it is wrong, what happened when you did otherwise, the rule. Example: "Everyone says respond to leads in five minutes. We waited a day and booked more calls."

2. Lesson from a failure

Shape: what you tried, how it failed (specific cost), what you misread, the changed rule. Example: "We spent four months building a feature nobody asked for. Here is the one question that would have saved us."

3. Before and after

Shape: the old state in numbers, the change, the new state, what mattered most. Example: "Our weekly report took six hours. Now it takes twenty minutes. One change did 80% of it."

4. The framework in steps

Shape: name the problem, give 3 to 5 steps that each carry a reason, end on which step people skip. Example: "How I decide whether to say yes to a speaking invite, in four questions."

5. The scene

Shape: open in a specific moment, the tension, what you did, what it shows. Example: "11 p.m., a customer emails to cancel and thanks us. That is when I learned what our product was really for."

6. Myth versus reality

Shape: "People think X. In practice Y." Two or three examples, then the practical consequence. Example: "People think senior engineers write the most code. The best ones I have hired delete the most."

7. What I would tell my younger self

Shape: a time marker, 3 to 5 pointed lessons each with a one-line reason, one thing you would still not change. Example: "Five years ago I priced too low out of fear. Here is what I would do in month one."

8. Observation from the field

Shape: a pattern seen across many cases (only if the person really has), what it suggests, what to do. Example: "In 30 hiring loops, the candidates who asked about failure got the offer. Why that tells you something."

Cómo añadir un skill

  1. Claude (web y escritorio): descarga el skill en .zip y añádelo en Ajustes, Capacidades, Skills.
  2. Claude Code: descomprímelo en ~/.claude/skills/ (o en .claude/skills/ de tu proyecto).
  3. Otros asistentes: abre el SKILL.md y pégalo en las instrucciones personalizadas o el conocimiento del proyecto.

Los skills están escritos en inglés. Tu asistente te responderá y escribirá los posts en tu idioma.