community-replies
Skill1 starsChanged 3 days agoMITView on GitHub
Find the Reddit and forum threads that shape AI answers where a helpful reply fits, and draft replies that state the writer's real role. Use for Reddit work for AI visibility. Works from a web search, or from the threads that a Frontmost AI project records. Drafts only. The person posts.
In this repository18 other files it gives its agents.
- ai-readinessSkill
- ai-search-strategySkill
- answer-accuracySkill
- answer-pagesSkill
- customer-surveySkill
- directory-listingsSkill
- listicle-outreachSkill
- mention-checkSkill
- name-checkSkill
- page-draftingSkill
- placement-trackerSkill
- question-draftingSkill
- question-gapsSkill
- rival-gapsSkill
- scan-reviewSkill
- site-auditSkill
- source-mapSkill
- tactic-checkSkill
The file
We split this skill into the rules it gives an agent
9 rules. About 0 are common, 8 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈8
Distinctive · estimate
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
- Distinctive: line 8
Use this guidance to plan replies in the Reddit threads that AI apps cite.
- Distinctive: line 12
Use the best source that you have.
- Very specific: line 20
If a Frontmost AI tool returns setup_required, give the person the url from the result.
- Distinctive: line 24
Keep a thread only when all of these are true:
- Distinctive: line 33
Ask who will post, and their role in the business.
- Distinctive: line 35
Read the subreddit's rules before you draft.
- Distinctive: line 44
No link unless the thread asks for one.
- Distinctive: line 45
Write it for one real person to post from their own account.
- Distinctive: line 60
Never invent a count, and never fill a gap with a guess.
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Community replies
Use this guidance to plan replies in the Reddit threads that AI apps cite. A reply helps only when it is useful to the people in the thread. A reply that reads as an ad gets removed and can get the account banned.
1. Find the threads
Use the best source that you have. Say which one you used.
| Source | How | Limit |
| A Frontmost AI project | Call get_sources for the project and the scan in the request. Read the subreddits, with threads_status and threads_needing_check. Open a subreddit with domain "reddit.com" and its subreddit name | Needs an account |
A source-map or a mention-check count | Take the Community rows, or the Reddit addresses in the Linked Pages column | A small sample |
| A web search | Search each buyer question with site:reddit.com. Take the threads in the first ten results | These threads stand in for AI citations. You have no citation counts |
If a Frontmost AI tool returns setup_required, give the person the url from the result. If it returns plan_required, say that this data needs a paid Frontmost AI plan. In both cases, continue with a web search.
2. Keep the threads that fit
Keep a thread only when all of these are true:
- AI apps cite it in this scan's period, or it is in the search results for a buyer question.
- The thread does not mention the business. A thread that needs a check is not confirmed: read it first.
- The question in the thread is one that the product really answers.
- The thread still takes replies. An archived or locked thread cannot. Say so, and leave it out.
3. Read the rules, then draft
Ask who will post, and their role in the business.
Read the subreddit's rules before you draft. If the rules forbid self-promotion, do not draft a reply for that subreddit. Say why. If the rules limit it, quote the limit and ask the person.
Draft one reply for each thread that remains.
The reply
- First line: the writer's real role, for example "I work at Acme" or "I am the founder of Acme".
- Answer the question first. The answer must help a reader who never tries the product.
- Mention the product once, where it fits the question. Say who it does not fit.
- No link unless the thread asks for one. No sales words.
- Write it for one real person to post from their own account.
Never suggest
- Fake accounts, more than one account, or coordinated posts.
- A reply that hides the writer's role.
- Requests for upvotes, or replies from employees who pretend to be customers.
- The same reply in more than one thread.
- A post in a thread where the product does not fit the question.
- Any promise that AI apps will mention the business after the reply.
Rules
- Frontmost AI tools do not open public pages. Read each page with your own web tool. If you cannot open a page, ask the person for its text, and say that you did not read it.
- Draft only. The person posts. Do not post for them.
- Never invent a count, and never fill a gap with a guess.
- Report each number with its counts and its source: AI answers, or web-search results. With a Frontmost AI project, link the
urlof the Sources tab. - Plan a small number of replies. Three good replies are better than ten that look the same.
Output
For each thread, up to 5: the thread, why it fits (the count with its source, and the question it asks), the subreddit rule that applies, and the draft reply. Then the threads that you left out, with the reason for each.
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.

