ai-citation-sources
Skill8 starsChanged 3 days agoMITView on GitHub
use this when AI answers (ChatGPT, Copilot, Gemini, Perplexity, AI Overviews) cite third-party pages instead of your site for prompts you care about, and you need to list those sources and pick a draft-only action for each one
In this repository28 other files it gives its agents.
- whetstone-signalSkill
- design-a-grok-botSkill
- make-bot-uiSkill
- overheard-setupSkill
- overheardSkill
- routine-healthcheckSkill
- transcript-healthcheckSkill
- design-taste-frontendSkill
- diagram-designSkill
- motion-pickerSkill
- redesign-existing-projectsSkill
- engineering-playbookSkill
- show-meSkill
- chatgpt-images-browser-playbookSkill
- aeo-geo-siteSkill
- ai-search-gap-loopSkill
- exact-match-intent-seoSkill
- markdown-twinSkill
- shorts-reframeSkill
- coachSkill
- draftSkill
- no-ai-slopSkill
- outlineSkill
- social-copywritingSkill
- source-checkSkill
- top-editSkill
- x-articleSkill
- youtube-niche-reportSkill
The file
We split this skill into the rules it gives an agent
9 rules. About 0 are common, 7 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈7
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 38
If a source does not fit any tag (for example a large reference site where a product mention would be self-promotion), write skip and the …
- Distinctive: line 45
Read each community's rules first.
- Distinctive: line 55
Prefer OWN and CLAIM first.
- Distinctive: line 57
Add the chosen action to the cycle entry in the ai-search-gap-loop log.
- Distinctive: line 66
No fake reviews, ratings, testimonials, or sock-puppet accounts.
- Distinctive: line 67
No astroturfing. Always disclose the connection to the product.
- Distinctive: line 68
No paid placement without clear disclosure on the page.
- Very specific: line 69
No invented numbers. Use missing when a value was not read.
- Distinctive: line 71
All outbound text is a draft.
≈ “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
AI citation sources
Use this skill for prompts where your site loses because AI answers trust other sources: lists, reviews, forums, directories, articles, video platforms. You list those sources. You tag each one OWN, CLAIM, EARN, or CONTEST. You write a draft action for each one. A human approves every action.
Input: the lost rows with cause third-party from ai-search-gap-loop. Each row has the prompt, the date, the tool, and the cited domains.
Helper file: source-table-template.md.
Step 1. List the sources
- For each lost prompt, copy the cited URLs in the order the answer gave them.
- Merge the URLs into one table. For each source, record:
- URL and domain
- Source type: list article, review site, forum or community, directory, video platform, news, reference, competitor
- Lost prompts that cite it (row IDs) and the count
- What it says about your category, and whether it names you
- Open each page. Read it. Do not tag a source from its title only.
- Sort by the number of lost prompts that cite it. Work from the top.
Step 2. Tag each source
Give each source exactly one tag.
| Tag | When | Draft action |
| OWN | You control the page (your site, your channel, your profile, your store page), but it is weak or not cited. | Draft an edit to that page so it answers the prompt plainly and matches the facts on your site. Use aeo-geo-site. |
| CLAIM | The source has a listing or profile slot for you that you can claim or fill as the owner: a directory, a platform profile, a knowledge panel, an app or catalog listing. | Draft the claim or profile text with true, consistent facts (name, what it is, who it is for, official links). Note the account and verification steps a human must do. |
| EARN | An independent editor, reviewer, or list author decides who is included. | Draft a short, factual note to the author or editor that says what the product is and why it fits their list. No payment for placement unless the page discloses it. No pressure. |
| CONTEST | The source has a factual error about you, or a fair comparison leaves you out of a set you clearly belong to. | Draft a correction request with evidence (a link to a public page that shows the fact). Ask for a correction, not removal of others. |
If a source does not fit any tag (for example a large reference site where a product mention would be self-promotion), write skip and the reason.
Step 3. Forums and communities
Forums such as Reddit, Quora, and topic boards are often cited. Treat them with care.
- Pick communities only when they are close to the product and to the buyer's decision. Use the tracked prompts to choose them, not community size.
- Read each community's rules first. Many ban self-promotion or links. Follow them exactly.
- Draft only. Never post, reply, vote, or message from this skill.
- Disclose the connection in every draft ("I make X"). Never pose as a neutral user, customer, or parent.
- Answer the question first. Mention your product only if it truly answers the question and the rules allow it.
- Respect any posting limits the account owner sets. Keep drafts few and specific.
- A human reads, edits, and decides whether to post each draft.
Step 4. Pick the work for this cycle
- Pick at most one or two source actions per cycle, so the loop can tell what helped.
- Prefer OWN and CLAIM first. You control them, and they are fast to check.
- EARN and CONTEST depend on other people. Log the date you sent them and do not count them as wins until the source changes.
- Add the chosen action to the cycle entry in the ai-search-gap-loop log.
Step 5. Recheck
On the next cycle, reopen each acted-on source. Record what changed on the page, with the date. Then rerun the lost prompts. Do not claim that a source change caused an AI change unless the rerun shows the source cited with your name in it.
Guardrails
- Same content for bots and people. No cloaking.
- No fake reviews, ratings, testimonials, or sock-puppet accounts.
- No astroturfing. Always disclose the connection to the product.
- No paid placement without clear disclosure on the page.
- No invented numbers. Use
missingwhen a value was not read. - On sites for children or families: no personal data about children in drafts, outreach, or tracking.
- All outbound text is a draft. A human approves before anything is sent or posted.
- Respect each site's terms and each community's rules.
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.

