agentsclimarketplace

Antwork ideas

Skill iker-gonzalez/antwork-skills/skills/antwork-ideas

Agent Skills for the Antwork MCP connector

Install
npx -y skills add iker-gonzalez/antwork-skills --skill antwork-ideas

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Use when the user wants content ideas, hooks, or angles for their social posts grounded in what has actually worked for their account — not generic brainstorming. Trigger on phrases like "give me post ideas", "what should I post", "I'm out of ideas", "content ideas for this week", "hooks for LinkedIn", "ideate", or any request for what to write next. Pulls the account's top performers and voice first, then generates ideas in that proven lane.

SKILL.md

2.9 KB, as published. Nobody here has run it

Generating content ideas in Antwork

Generic brainstorming is worthless — the user can get "10 LinkedIn post ideas" from any chatbot. Antwork's edge is data: this account's real top performers and its captured voice. Ground every idea in both. Ideas should look like "more of what already worked for you," not a listicle.

1. Pull the evidence first — never ideate cold

Before generating anything, gather:

  1. get_performance(limit) — the account's top posts by lifetime engagement. This is the proven lane.
  2. get_post_context(platform, account_id) — the voice profile, brand context, and recent 3 posts. If voiceStale is true, hand to antwork-voice first; ideas in the wrong voice waste everyone's time.

If the account is new and has no performance history, say so plainly and ideate from voice + brand + the user's stated goals instead — but don't pretend you're being data-driven when you aren't.

2. Find the pattern, then extend it

Read the top performers and name why they worked — hook type (bold claim, contrarian take, question, personal story, useful list), topic cluster, format, length. Then generate ideas that reuse the winning mechanism on fresh material. "Your contrarian one-liners outperform — here are 6 more contrarian takes on adjacent topics" beats a random grab bag.

Mix the slate so it isn't one-note: ~60% in the proven lane, ~40% adjacent experiments to find the next winner.

3. Output a ranked, decision-ready list

Give a numbered list. Each idea is one row:

  • Hook — the actual opening line, written in the account's voice (not a topic label).
  • Platform — the account it fits, respecting that platform's norms.
  • Angle — the one-sentence shape of the post.
  • Why — the evidence: which past winner or voice trait it draws on.

Rank by expected payoff (proven-lane ideas first). Keep hooks tight and in-voice — apply the same anti-AI-tells rule the voice profile enforces (no "Here's the thing:", no "Let me break it down", no emoji-spray openers).

4. Hand off cleanly

Ideas are upstream of action. Close by offering the next step:

  • Turn a pick into a draft → antwork-poster.
  • Spread a batch across the week → antwork-calendar.
  • Build a themed sequence → antwork-campaign.

Don't call create_post from here unless the user picks an idea and says go — ideation produces a menu, not drafts.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.