For the complete documentation index, see llms.txt. Every page on this site is also served as Markdown: append `.md` to any URL, or send `Accept: text/markdown`.
Skip to content
Attensira LogoAttensira
Start free

What to ask your assistant about your AI visibility

One page per question people ask on a Monday morning, answered through Attensira's MCP server from Claude, ChatGPT, Cursor or any MCP client.

Each use case is one question, such as where you lost ground this week or which sources the models trust, with the single Attensira MCP call that mostly answers it, the chain behind that call, how to read the result, and what the numbers cannot tell you. Every example is read from our own workspace.

7 days free · card requiredHow the trial works

ChatGPTattensira · attensira.com

Asked: Which of our tracked questions do AI answers name us on most?

called get_analytics area=prompt days=30 order=best_first limit=4

Only the questions that already contain our name. The next one down, a category question, names us in none of its answers.

"what is Attensira"
60 of 60
"Attensira vs Profound"
60 of 60
"Attensira reviews"
60 of 60
"AI crawler logs analysis tool"
0 of 105
Ahrefs named in 18 of 105

The honest starting point: we win our own name and little else yet.

attensira.com workspace, read over the Attensira MCP ·

Which question are you asking?

Each page gives the call to paste, the chain behind it and the limits of what comes back.

What does one of these reads return?

The first call of “See which sources AI answers lean on”, get_analytics area=source, on our own workspace: every domain the answers cited in 30 days, each over all 105,376 citations. Our own site is seventh; the six above it are the work list.

get_analytics · sourcesn=105,376 citations
ChatGPT and Perplexity, every tracked prompt · to
Nameget_analytics · sources
help.openai.com5.3%
developers.google.com4.4%
semrush.com2.3%
otterly.ai2.2%
peec.ai2.1%
openai.com1.5%
attensira.comyou1.5%
attensira.com workspace, read over the Attensira MCP · · n=105,376 citations

What do I need before I start?

A connected client

Claude, ChatGPT, Cursor, VS Code, Claude Code, Codex CLI, Zed or n8n, connected to mcp.attensira.com/mcp.

Tracked prompts

Every number is read from the questions your workspace tracks. A competitor you have not added reads as absent, not as losing.

Enough answers

Sampling is at most once per prompt per engine per day on Starter, three times on Growth and up. A rate built on single-digit n swings for reasons that are not yours.

What else do people ask?

Do I need to write the tool calls myself?
No. Ask your assistant the question in your own words and it picks the call. Each page shows the exact call so you can check what it did, paste it into a client that takes raw tool calls, or put it in an automation.
Which of these use cases spend credits?
Only the steps that start agent work: ask, run_automation and add_prompts. Every read in these chains, including win_plan, is free however often you call it. Each page says whether its chain spends credits.
Why does every page have a section on what it cannot tell you?
Because a number read without its limits overstates the product. Short windows on few answers swing, a competitor you do not track reads as absent, and a citation is not traffic. The limits are part of the answer.
Can an automation run these questions for me every week?
Yes. create_automation takes a plain-language instruction, a trigger and a time, and runs it on schedule. Creating one is free; each run spends credits like a manual one. The weekly digest use case shows the instruction we would write.

Put the crew to work on your brand.

7 days free · card requiredHow the trial works

Start free

7 days free · card requiredHow the trial works

attensira.com's own workspace · read 4 Oct 2026 · n=4,518

197/ 4,518 answers4.4%

Answers on ChatGPT and Perplexity that named attensira.com. Our own reading, published with its n.Sites we own · not customers · 24 of 31 days sampled