mcpbeat Sign in

Twitter Post Agent Skill

by langchain-ai

Write a Twitter/X post or thread based on research findings or a given topic. Use this skill when asked to create tweets, X posts, or social media threads.

587 tokens
context cost
the whole folder, loaded on every use
1
files
instructions only
0
copies elsewhere
how many repositories repackaged it
15
stars on the repo
on the repository, not the skill itself

Install

one command, takes just this skill from the repository 
npx skills add https://github.com/langchain-ai/langgraph-101-ts --skill twitter-post

What it tells the agent to use

found in the instruction text
Task spawns other agents

The instruction itself

7 sections, as written by the author

Twitter/X Post Skill

Single Tweet Format

  • Maximum 280 characters
  • Lead with the most compelling point
  • Use numbers or data when possible
  • End with a link placeholder or call-to-action
  • 1-2 hashtags max (optional)

Thread Format (for longer content)

  • Tweet 1: Hook + preview of what's coming (e.g., "A thread on X:" or "Here's what I found:")
  • Tweets 2-N: One idea per tweet, numbered (1/, 2/, 3/)
  • Final tweet: Summary + call-to-action + link
  • Keep each tweet self-contained (people share individual tweets)
  • 4-8 tweets is the sweet spot for engagement

Tone

  • Concise and punchy
  • Opinionated takes perform better than neutral summaries
  • Use plain language -- no corporate speak
  • Contrarian or surprising angles get more engagement

Tips

  • Front-load the value (no throat-clearing or preambles)
  • Use line breaks within tweets for readability
  • Avoid hashtags in threads (they look spammy) -- save them for single tweets
  • Numbers and lists catch the eye in a feed

Example Single Tweet

AI agents that manage their context window well outperform those with 10x more tools.

The secret isn't more capabilities -- it's smarter context engineering.

Example Thread

Thread: What makes AI agents actually work in production? 🧵

1/ It's not the model size. It's context management.

The best agents treat their context window like RAM -- offloading to filesystem, summarizing aggressively, loading info on demand.

2/ Subagents are the key to scaling.

Instead of one agent doing everything, delegate to specialists. The main agent only sees the summary, not 50 intermediate tool calls.

3/ Skills > giant system prompts.

Progressive disclosure: load detailed instructions only when the task needs them. Your agent's prompt stays clean until it matters.

4/ Memory needs structure.

Semantic (facts), episodic (experiences), procedural (rules) -- route them to different backends so they persist appropriately.

5/ The takeaway: the best agent architectures are about information flow, not raw capability.

What patterns are you using? Reply with your favorite agent architecture trick.

How to use it

Copy the folder

Take langchain-ai/langgraph-101-ts-twitter-post from the repository into ~/.claude/skills for personal use, or into .claude/skills inside a project.

Check the name does not clash

The agent identifies a skill by the name field in its header. Two skills with the same name cannot sit side by side — one of them will be ignored.