x-reply
Skill1 starsChanged 6 months agoMITView on GitHub
Build a reply-based growth strategy for a new or small X (Twitter) account, and draft high-quality replies. Use when the user has few followers, asks "how do I grow on X from zero", "my account is new", "who should I reply to", "help me write a reply to this tweet", or wants a cold-start plan. Returns an anchor-account strategy and concrete reply drafts.
3 other projects use a version of this.
Adopted by
Most recently taken up by
- belzoralolbtw/x-agent-skilltheir own version, first seen Oct 25, 2026
- swayamdhawan/x-agenttheir own version, first seen Oct 20, 2026
- Jakeschincariol/x-agent-skilltheir own version, first seen Sep 29, 2026
Other versions3 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 3
- Jakeschincariol/x-agent-skill★ 4same name
- belzoralolbtw/x-agent-skill★ 0same name
- swayamdhawan/x-agent★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository5 other files it gives its agents.
- x-auditSkill
- x-hookSkill
- x-profileSkill
- x-teardownSkill
- x-threadSkill
The file
We split this skill into the rules it gives an agent
8 rules. 0 are common, 7 are distinctive, and 1 is very specific to this project.
0
Common
Most of the 3 edited versions we read keep these.
7
Distinctive
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 8
Read reference/ALGORITHM.md if available.
- Distinctive: line 19
When those people like/follow you, your author embedding gets pulled toward that cluster — which is what makes your *own posts* start …
- Distinctive: line 26
When the user wants a plan.
- Distinctive: line 35
Coherence: all replies should stay inside ONE niche.
- Distinctive: line 40
When the user pastes a tweet they want to reply to.
- Distinctive: line 73
Never invent X handles. If you don't know real accounts in the niche, ask the user.
- Distinctive: line 74
Never write filler replies.
- Distinctive: line 76
Keep replies short. Most strong replies are 1–3 sentences.
“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
x-reply — Reply Growth Engine (Cold Start)
Read reference/ALGORITHM.md if available. This skill solves the hardest problem on X:
a new account with no followers has almost no distribution.
Why. Two facts from the open-source code:
- The out-of-network scorer multiplies your reach by a factor below 1 — in-network is structurally favored.
- Out-of-network retrieval needs your author embedding to be known and well-placed. A new account's embedding is cold — the model doesn't yet know which cluster you belong to.
Replying under the right accounts is the fix. It does two things at once:
- Puts you directly in front of an audience already engaged with your exact topic.
- When those people like/follow you, your author embedding gets pulled toward that cluster — which is what makes your own posts start getting retrieved later.
Replies are not a side activity for a new account. They are the main activity.
Mode A — build the anchor-account strategy
When the user wants a plan. You need their niche/topic and what they post about.
- Identify 10–15 anchor accounts in their exact niche. Prefer accounts with 5k–200k followers (big enough for reach, small enough that replies are seen) over mega-accounts where replies drown. If you don't know real accounts in their niche, ask the user to name a few they already follow and build from there — do not invent handles.
- Set a cadence: ~70% replying, ~30% posting for the first 30 days. Roughly 20–50 thoughtful replies/day, concentrated within the first 15–30 minutes of an anchor's post (early replies get seen).
- Coherence: all replies should stay inside ONE niche. A scattered reply history blurs the author embedding just like scattered posts do.
Mode B — draft a specific reply
When the user pastes a tweet they want to reply to. A reply that grows you must:
- Add something — a concrete data point, a counter-example, a sharper version of the idea, a real experience. Never "Great post!", "This 💯", "So true".
- Stand alone — it should make sense and be interesting to someone who reads only the reply, because that's how strangers discover you.
- Be entity-dense — a number, a tool, a name. Same embedding logic as posts.
- Not hijack — don't pitch your product under an unrelated post; that trains
not_interestedand reads as spam togrox. - Invite continuation — a reply that earns a response from the original author or other repliers gets far more eyes.
Output format
For Mode A:
Niche: <understood niche>
Anchor accounts (10–15): <list, grouped by sub-topic if useful>
Daily routine: <reply count, posting ratio, timing windows>
First-week focus: <what to do days 1–7>For Mode B:
Original post: <one-line summary>
Reply drafts (best first):
1. "<reply>" — <why it works: which signal it triggers>
2. ...
Recommended: #<n>Rules
- Never invent X handles. If you don't know real accounts in the niche, ask the user.
- Never write filler replies. A filler reply is worse than no reply — it spends the anchor's audience attention and earns nothing.
- Keep replies short. Most strong replies are 1–3 sentences.
- Be honest: cold start is slow. Set the expectation that this is weeks of work, not days.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

