The article pipeline is within reach: search the keyword, read the top pages, outline, write section by section with sources attached, then drop a draft into WordPress. An agent can build that in a weekend on your own LLM key and a search API. KoalaWriter's edge is the data behind the loop: SERP analysis, live Amazon product data, keyword and backlink lookups and an index of your site for internal links, all bought or maintained for you. Reviewers also warn that articles need fact-checking and editing before they go live, and a Trustpilot reviewer says sites filled with its output ranked worse than ones written with ChatGPT.
Before you buildKoalaWriter gives 5,000 free words and 25 credits once. Write two or three articles and compare them with the output of the pipeline above before paying.
Build me a local SEO article writer to replace KoalaWriter. Requirements: - Node.js 22 + TypeScript: a Fastify server and a React UI built with Vite. `npm start` serves http://localhost:4800, and `npm run write -- "<keyword>"` runs the same pipeline from the shell. LLM: any OpenAI-compatible endpoint through the `openai` SDK (LLM_BASE_URL, LLM_API_KEY, LLM_MODEL in .env), with progress streamed to the UI over server-sent events. - Research: call the search API chosen in .env (Brave Search or Tavily) for the keyword, fetch the top 8 pages, extract text with jsdom + @mozilla/readability, and record headings, word counts and the terms that at least 4 pages share. Writing: outline first, then each section with the shared terms, the extracted facts and my voice notes. Every factual sentence keeps a [n] marker that maps to a source URL stored with the article. - Storage: better-sqlite3 in ./data/koala.db: sites (name, base URL, sitemap URL, voice notes), keywords, articles (Markdown, HTML, status, word count, sources) and an FTS5 index of site pages. - Internal links: read the site's sitemap.xml, index titles and H1s, and add 3 to 6 links per article using anchor text already in the draft. Never invent a URL. - Bulk mode: paste up to 50 keywords or upload a CSV; concurrency from .env (default 2) and a token cap per run that stops the queue when reached. - Checks before an article is "ready": each [n] resolves to a source, no placeholder text, word count within 15% of target, keyword in the title and first paragraph. Failures stay in "needs review" with the reason shown. - Publish: a WordPress REST draft through an application password (draft only), or download as Markdown or HTML. An optional featured image from an OpenAI-compatible images endpoint. - Single user. No accounts, no telemetry, local-first: only the LLM, the search API, the pages fetched and my WordPress site receive requests. - Out of scope: Amazon product data, keyword volumes, Search Console, lead magnets, teams. - README: the .env keys, what leaves the machine, token counts printed after each run, and a warning to read and edit every article before it is published.
Opens your agent with the prompt filled in, you press enter. Or copy it and paste it into an empty folder.
The first prompt builds the core. Run these one at a time in the same folder to grow it into your own KoalaWriter.
Extend the article writer in this folder. Requirements: - Add a refresh mode: give it a published URL and a keyword. Fetch the page, re-run the research step, and list sections that are missing terms the top pages share. - Rewrite only the flagged sections, keep the headings, URL slug and existing internal links. - Show a side-by-side diff in the UI and save the result as a new version, never overwriting. - Push the approved version as a WordPress draft update only after I confirm. - Update the README.
Add a keyword picker to the article writer in this folder. Requirements: - Import a Search Console performance CSV (query, page, clicks, impressions, position). - List queries at positions 8 to 30 with at least 50 impressions, grouped by shared words and by the page that currently ranks. - Mark clusters that have no matching article in the local database and let me send a cluster to the bulk queue in one click. - Everything stays local; no Google login. - Update the README.
Add an image pass to the article writer in this folder. Requirements: - After an article passes checks, propose one image prompt per H2 and show them for approval. - Generate approved images through the OpenAI-compatible images endpoint from .env, save them to ./data/images/, and insert them with alt text written from the section's first sentence. - Cap images per article in .env and print the count before generating. - Include the images when publishing to WordPress as media uploads. - Update the README.
Content sites pay to skip the plumbing: one box takes a keyword and returns a structured draft with sources, internal links and images, at a fixed monthly fee instead of a pile of API bills. For someone publishing dozens of pages a month, the bulk mode and the connected research tools save more than $25 of attention.
Estimate · Running cost is model tokens plus search API queries, so it scales with the number of articles; a few dozen a month will often undercut $25. The build is token usage on your agent plan.
| Plan | Monthly | Annual, per month | What you get |
|---|---|---|---|
| Free trial | $0 | $0 | 5,000 free words and 25 credits, one time. |
| Starter | $25 | $20 | 45,000 words and 500 credits a month, 1 Brand DNA brand, bulk writing, real-time data, AI images, Amazon data for affiliate articles, WordPress, Shopify, Webflow, Ghost and webhook publishing, API and MCP access. |
| Professional | $49 | $39 | 100,000 words and 1,000 credits a month, Search Console and Analytics dashboard, article editor with SEO score, KoalaLinks for up to 50 sites, KoalaMagnets, Deep Research. |
| Boost | $99 | $79 | 250,000 words and 2,500 credits a month, full Brand DNA experience, uncapped Brand DNA agent, Content Calendar Autopilot, 2x faster bulk creation. |
| Growth | $179 | $143 | 500,000 words and 5,000 credits a month, 3 brands. |
| Elite | $350 | $280 | 1,000,000 words and 10,000 credits a month, 7 brands. |
| Advanced | $500 | $400 | 1,500,000 words and 12,500 credits a month, 12 brands. |
| Scale I to III | $750 | $600 | 2.5M, 5M or 10M words and 15,000 to 25,000 credits a month, 20 to 75 brands; 3x faster bulk creation. |
A new account gets 5,000 free words and 25 credits with no card; free accounts can't connect an AI assistant over MCP.
Kinda. The core is a weekend build with the prompt on this page, but real gaps remain: Keyword volumes, backlink and competitor data included in the plan, Live Amazon product data for affiliate roundups.
About $25 a month on the Starter plan, $300 a year (checked 2026-10-19).
Keyword volumes, backlink and competitor data included in the plan; Live Amazon product data for affiliate roundups; Search Console and Analytics dashboards with an SEO agent on top; An editor that scores your draft against the pages that rank; One-click publishing to Shopify, Webflow and Ghost as well as WordPress. If any of those matter to you, keep paying.
Yes: GPT Researcher (open source), Dify (open source).