AgentSkillsCN

commenting

制定跨平台(X、LinkedIn 等)与他人互动的策略。寻找目标对象,撰写有价值的回复,建立联系。

SKILL.md
--- frontmatter
name: commenting
description: Strategy for engaging with others' posts across platforms (X, LinkedIn, etc.). Finding targets, writing valuable replies, building connections.
user-invocable: false

Commenting / Engagement Skill

Write valuable replies, engage with others

Commenting is how you turn visibility into relationships. Publishing broadcasts; commenting connects.

Why Commenting Matters

  • Exposure: replies to larger accounts put you in front of their audience
  • Algorithm boost: reply-to-reply = 75x multiplier on X
  • Relationships: genuine engagement builds real connections
  • Authority: thoughtful replies showcase expertise without self-promotion
  • Reciprocity: people check profiles of good commenters and follow back

Finding Reply Targets

Use the discovery skill to find posts worth replying to. Targets are stored in agent/memory/research/reply-targets.md.

Writing Good Comments

The Value Test

Before posting a reply, ask: would someone follow me based on this reply alone?

What Works

PatternExample
Add data/evidence"We tested this — saw 3x improvement when..."
Build on the idea"Great point. I'd add that [insight]..."
Share experience"Ran into this exact problem. What worked was..."
Ask a sharp question"Curious about X — have you seen it work at scale?"
Respectful disagreement"Interesting take. I've seen the opposite because..."
Connect dots"This pairs well with [other person]'s point about..."

What Doesn't Work

  • Empty agreement ("Great post!", "So true!", "This.")
  • Self-promotion without context ("Check out my thing!")
  • Arguing for the sake of arguing
  • Generic responses that could apply to any post
  • Long essays in reply (save those for your own posts)

Voice

  • More conversational than publishing voice
  • Shorter — get to the point fast
  • Show curiosity — ask genuine follow-up questions
  • Be specific — reference what they actually said
  • Credit freely — "building on your point..."

Reply File Format

X (Twitter)

File: agent/outputs/x/reply-YYYYMMDD-NNN.txt

code
REPLY_TO: 2019637612076494985
---
Your reply text here.

Tracking Targets

Store found targets in agent/memory/research/reply-targets.md:

markdown
# Reply Targets
Last updated: YYYY-MM-DD

## Pending
- ID: 2019637612076494985 | @username | "Post summary..." | Reply angle: [your insight]

## Replied
- ID: ... | @username | Replied YYYY-MM-DD

Allocation

Target 20-30% of output as replies (not just original posts):

  • 2-3 replies per session alongside 1-2 original tweets
  • Don't batch — spread across sessions for natural engagement
  • Always engage with people who reply to YOUR posts (reply-to-reply)

Platform Notes

X

  • Reply-to-reply = 75x algorithm multiplier
  • Early replies get more visibility (within first 30 min)
  • Thread replies (replying multiple times) can work but don't overdo

LinkedIn (future)

  • Comments on posts boost your visibility in the commenter's network
  • Longer, more thoughtful comments perform better than on X
  • Tagging the author in your comment is implicit (it's their post)