Intermediate 📅 Last Updated: July 1, 2026 ⏱️ 12 min read 🪿 Goose

How to Use Goose to Research Blog Topics and Build a Content Plan

Who This Is For

Read this if: You write a blog, run a content site, or build documentation and want a local, private AI workflow for topic research and planning — no data sent to OpenAI.

Skip if: You have not installed Goose yet. Start with the Goose guides and come back.

What You Need

• Goose installed and configured with a local model (see Goose setup)
• A project folder where Goose can write its research files
• A niche or topic area you want to write about
• Time: 30 minutes for a full content plan

How This Guide Was Made (Yes — Really)

Here is the meta part: this very article was researched and outlined by Goose. The workflow below is not theoretical. It is the exact process used to plan MasterAIKit's launch content. The prompt examples are real prompts from the build.

The 4-Stage Research Workflow

Step-by-Step Guide

Step 1: Set Up the Project Folder

mkdir -p ~/content-research/outlines
cd ~/content-research

# Start Goose in this directory with write access
goose session --path . --permissions read write

Step 2: Stage 1 — Generate Topic Ideas

Give Goose your niche and audience. Be specific — vague prompts produce generic slop.

Prompt (paste into Goose):

I run a blog called MasterAIKit about running AI locally
on your own hardware. Audience: developers and hobbyists
who want privacy, no API costs, and offline AI.

Generate 30 blog topic ideas. For each, include:
- A working title (SEO-friendly, not clickbait)
- The target keyword
- The user intent (informational / comparison / tutorial)
- One sentence on why a reader would search this

Save the list to topic-ideas.md. Rank the top 10.

Goose writes the file directly. Open topic-ideas.md and review — you will get 30 concrete ideas, not "write about AI."

Step 3: Stage 2 — Analyze and Score

Now ask Goose to score the ideas so you prioritize the right ones.

Prompt:

Take the 30 ideas in topic-ideas.md. Score each on:
1. Search intent clarity (1-5) — would someone actually Google this?
2. Competition difficulty (1-5) — is the SERP dominated by big sites?
3. Content gap (1-5) — does existing coverage miss something?
4. Monetization fit (1-5) — does it tie to our Starter Kit or Review?

Add a total score. Sort descending. Save to topic-scoring.md.
Highlight the top 10 in bold.

Step 4: Stage 3 — Build Full Outlines

This is where the workflow gets powerful. Goose turns each top idea into a complete outline you can hand to a writer (or write yourself).

Prompt:

Take the top 10 topics from topic-scoring.md. For each,
write a full article outline and save to outlines/[slug].md.
Each outline must include:

- Target keyword and 2 secondary keywords
- Word count target (800 / 1200 / 1500)
- H1 and all H2/H3 headings
- A 2-sentence "quick answer" for the intro
- 3 internal linking opportunities to other outlines
- 1 CTA suggestion (Starter Kit / Setup Review / Free Checklist)
- The "who this is for / skip if" framing

Create one file per topic in the outlines/ folder.

You now have 10 ready-to-write outlines. Each one is a draft structure, not a blank page.

Step 5: Stage 4 — Build the Content Calendar

Prompt:

Review all outlines in the outlines/ folder. Build a
4-week content calendar in content-calendar.md.

Rules:
- 3 articles per week (Mon, Wed, Fri)
- Sequence by topic clusters (group related guides together)
- Week 1 = beginner funnel, Week 4 = advanced
- Note which articles link to which (internal linking map)
- Mark pillar articles that other guides link back to

Sample Output — From This Site's Real Plan

Here is an actual excerpt from MasterAIKit's topic-scoring.md that Goose generated:

| Topic                          | Intent    | Diff | Gap | Total |
|--------------------------------|-----------|------|-----|-------|
| How Much VRAM for Local AI     | Info      | 3    | 5   | 18    |
| Install Ollama Run First Model | Tutorial  | 2    | 4   | 17    |
| Best Local AI Models Beginners | Comparison| 4    | 3   | 16    |
| Give AI Agent File Access Safe | Tutorial  | 2    | 5   | 16    |
| Use Obsidian as AI Memory      | Tutorial  | 1    | 5   | 15    |

Those top topics became the guides you see in the MasterAIKit sidebar. The workflow works.

Common Mistakes & Errors

Mistake 1: Vague Niche Prompts

Cause: "Write about AI" gives you 30 useless ideas. Goose mirrors the vagueness back.

Fix: Specify the audience, the pain point, and the angle. "Developers who want private AI on a 12GB GPU" beats "AI blog ideas" every time.

Mistake 2: Skipping the Scoring Stage

Cause: You jump from ideas to writing and pick topics by gut feel.

Fix: Always run Stage 2. The scoring surfaces high-gap topics you would otherwise miss — like the agent file-access guide that scored 16 because competition was low but intent was high.

Mistake 3: Letting Goose Write the Final Article

Cause: Outlines are good, so you ask it to write the full article. The output is generic.

Fix: Goose plans; you write. Use the outline as a skeleton and add your tested-on data, real commands, and personal voice. That is what makes MasterAIKit content rank — humans add the testing layer.

Mistake 4: Not Saving Intermediate Files

Cause: You do the whole workflow in one chat session and lose it when the session ends.

Fix: Each stage writes to a file. That is intentional. You can re-run Stage 3 next week without redoing Stages 1 and 2.

Recommended Setup (Tiered)

What I Would Do

Run the full 4-stage workflow once to launch a site. Then each week, re-run only Stages 1–2 for fresh ideas and Stage 3 for the next batch of outlines. Keep content-calendar.md as your single source of truth — when an article is published, mark it done in the calendar. This is the exact system that built MasterAIKit's first 15 guides in a week.

Frequently Asked Questions

Can Goose write complete blog articles?

Goose can draft full articles by researching, outlining, and writing sections, but output needs human editing for voice, accuracy, and flow. It works best with a detailed brief and reference articles. Use Goose for first drafts, then refine.

How good is Goose at researching blog topics?

Goose is strong at gathering information, summarizing competing content, and identifying gaps. With web scraping and search MCP tools, it analyzes top-ranking articles and suggests angles. Quality depends on model choice.

Does Goose fact-check automatically?

No, Goose does not auto-fact-check. You must verify claims, statistics, and citations. AI models can hallucinate facts. Always cross-reference key data against primary sources before publishing.

What is the best content workflow?

Have Goose research and outline first, then draft each section while you review. Finally, ask Goose to check SEO keywords and generate meta descriptions. This iterative approach produces better content than one-shot generation.

Can Goose integrate with Obsidian?

Yes, Goose reads and writes to Obsidian vaults since they are markdown folders. Point Goose at your vault directory to organize research notes, create articles from templates, and maintain content calendars via the filesystem MCP.

Last Updated: July 1, 2026 — Verified against Goose 1.0 with qwen2.5:14b on Ubuntu 26.04 LTS.