You are an SEO content strategist. Target keyword: [keyword].
Here are the H1-H3 headings and word counts of the top 10 ranking pages: [paste].
Here are the People Also Ask questions: [paste].

1. State the dominant search intent and the content format that wins this SERP.
2. List the subtopics every top page covers.
3. List useful subtopics none of them cover well.
4. Write an H2/H3 outline that covers the must-haves plus at least two gaps.
5. List entities and terms to include, questions to answer in 40-60 word snippets, and 3 internal link targets from this list: [URLs].
Format it as a brief a freelance writer can follow without asking questions.

Here is a CSV of keywords with monthly search volume and the top 10 URLs ranking for each: [paste].

Group keywords that share 3 or more ranking URLs into the same cluster. For each cluster return:
- a parent topic name
- the primary keyword (highest volume)
- search intent (informational, commercial, transactional or navigational)
- total cluster volume
- which of our URLs already targets it, from this list: [our URLs], or "new page"

Return a table. Use only keywords and numbers from the data. Do not invent volumes or keywords.

Using the brief below, draft only the section titled "[H2]".
Write in the voice described in my style guide: [paste].
Use only facts from these sources: [paste notes, data, quotes].
Where a claim needs a source I haven't given you, write [NEEDS SOURCE] instead of inventing one.
Open with a direct one-sentence answer, keep paragraphs to 4 sentences or fewer, and include one concrete example.

Brief: [paste]

Here is this month's Search Console data compared with last month, by page and query: [paste CSV].
Here is our work log for the month: [paste].
Known outside events: [algorithm updates, seasonality, site changes].

Write a client update with:
1. Three headline results, each with a number.
2. The three biggest drops and the most likely cause for each, based only on the data and the log.
3. What we shipped.
4. Next month's three priorities.
Use plain language, stay under 300 words, and flag any conclusion the data can't support.

Here is the current text of our page [URL]: [paste].
Here are the headings of the 5 pages now outranking it: [paste].
Its clicks fell [X]% year over year.

List:
1. Statements or statistics that look outdated (quote them).
2. Subtopics competitors cover that we don't.
3. Questions from this People Also Ask list we don't answer: [paste].
4. A prioritized update plan, biggest impact first.
Don't rewrite the page yet.

Below are the titles, meta descriptions, headings and publish dates of the top 10 results for [keyword], plus the SERP features shown (snapshot from [date], [location]): [paste].

Describe:
1. The content format Google rewards here.
2. The angle most results share.
3. How fresh the results are.
4. Which result is weakest and why.
5. What a new page would need to stand out.

For each row (url, current title, primary keyword, page summary), write:
- a title tag under 60 characters that leads with the primary keyword and ends with " | [Brand]" where it fits
- a meta description of 140 to 155 characters with one clear benefit and a call to action

Don't repeat the same opening across rows. Return CSV with the columns url,title,meta.

Rows: [paste]

Generate JSON-LD for this page using the schema.org types Article and Organization.
Use only these facts: [headline, author name and URL, datePublished, dateModified, image URL, publisher name, logo URL, sameAs profile URLs].
Do not add any property I haven't given a value for.

Output a single script tag, then list recommended properties that are missing so I can supply them.

Here is a Screaming Frog export of issues with URL counts and sample URLs: [paste].

1. Group the issues by root cause.
2. Rank the groups by likely impact on crawling, indexing and rankings.
3. For the top 5, write a developer ticket with: problem, affected URL pattern, recommended fix, and how to verify the fix.
Mark anything you're unsure about instead of guessing.

Here are a journalist's last 5 article headlines and summaries: [paste].
Here is our new data study: [summary with the 3 strongest findings].

Suggest 3 story angles that fit what this journalist covers, then write a 90-word pitch for the best one.
Lead with the finding. No flattery, no "I hope this finds you well," and no claims beyond the study summary.

Draft replies to these 5 Google reviews for [business name]: [paste reviews].

Thank each reviewer by first name and mention the specific service they named.
Keep each reply under 60 words.
For negative reviews, acknowledge the issue and offer direct contact at [phone or email].
Don't promise refunds or make claims about the business that aren't in this fact sheet: [paste].

Build a Python script that takes a keyword and a Google Doc URL.
1. Call the DataForSEO SERP API (credentials in .env) for the top 10 organic results in the US.
2. Download each result and extract the main text with trafilatura.
3. Send the combined competitor text to the Claude API and ask for the 25 most important subtopics and entities, with how many pages cover each.
4. Read my draft from the Google Doc and mark each subtopic as covered, partly covered or missing.
5. Write the results to a Google Sheet with one row per subtopic.
Add error handling for pages that fail to download, and keep API calls under 20 per run.

Write a Google Apps Script that runs on the 2nd of each month.
1. Pull last month's and the prior month's Search Console data for [site] by page and query, using the Search Console API.
2. Write both to a sheet named "GSC Raw" and build a "Movers" tab with the 20 biggest click gains and losses by page.
3. Send the Movers tab as CSV to the Claude API with this instruction: "Write a 200-word client summary. Lead with results, explain the top three drops using only the data, and list next steps."
4. Email the summary and a link to the sheet to [address].
Include setup steps for the OAuth scopes I need.

Build a weekly rank tracker in Python.
Input: keywords.csv with keyword, target_url, location_code, device.
For each row, call the DataForSEO Google Organic SERP API and record: date, keyword, our position (match by domain [domain]), ranking URL, whether an AI Overview is present, and whether our domain is cited in it.
Append results to a SQLite database, then output a CSV of keywords that moved 3 or more positions since last week.
Add a dry-run mode that estimates API cost before running.

Write a Python script that compares two Screaming Frog "Internal - All" CSV exports (last week and this week).
Report:
1. URLs that changed from indexable to non-indexable.
2. Canonicals that now point to a different URL.
3. New 4xx and 5xx URLs with their inlink counts.
4. Title or H1 changes on URLs that contain /product/ or /services/.
Output a markdown summary sorted by severity, then a CSV with every flagged URL.

Build an AI visibility tracker in Python.
Input: prompts.csv (prompt, topic, stage) and brands.csv (our brand plus 5 competitors with domains).
For each prompt, call the OpenAI and Anthropic APIs with web search enabled, 3 runs each.
Save the full answer and every cited URL.
Then run a second pass that records, for each answer: which brands are mentioned, in what order, with what sentiment, and which domains are cited.
Output a weekly CSV with share of voice by topic and platform.