Workflows

Ready-to-use AI workflows

Each workflow is a few prompts, agents and tools wired together. Pick one, fill in the inputs, and run it in the agent you already use.

Content workflow

SEO article from a keyword

Researches the search results, outlines, drafts and then grades the draft against a rubric until it passes.

Runs in: Claude Code · Codex · ChatGPT · Claude.ai

With your inputs:Open in ClaudeOpen in ChatGPT

  1. TriggerManual start
  2. Tool / MCPSearch the top results
  3. PromptOutline
  4. PromptDraft
  5. Review loopGrade and revise
  6. OutputSave as Markdown
Fill in the inputs

Save as .claude/skills/seo-article-from-keyword/SKILL.md for Claude Code, or .agents/skills/seo-article-from-keyword/SKILL.md for Codex and other Agent Skills tools. Then ask: “run the SEO article from a keyword workflow”.

SKILL.md
---
name: seo-article-from-keyword
description: Researches the search results, outlines, drafts and then grades the draft against a rubric until it passes. Use when asked to run the "SEO article from a keyword" workflow.
---

# SEO article from a keyword

Researches the search results, outlines, drafts and then grades the draft against a rubric until it passes.

## Inputs

Ask for any of these that the user hasn't given:

- `keyword` (e.g. "best standing desk for small apartments")
- `audience` (e.g. "renters working from home")

## Steps

Run the steps in order. Refer to saved results by name. Don't skip a step that says STOP.

1. **Manual start** — Start: Start with {{keyword}} and {{audience}}.
2. **Search the top results** — Use web_search: Search for {{keyword}}. Collect the top 8 results: title, headings, angle and what each one misses. Save the result as `serp`.
3. **Outline** — Using {{serp}}, write an outline for {{audience}} that covers the gap the results miss. H2s only, one line each. Save the result as `outline`.
4. **Draft** — Write the article from {{outline}}. 1,200 words, plain language, one table comparing the desks. Save the result as `draft`.
5. **Grade and revise** — Review loop (max 3 passes): Score {{draft}} 1–5 on: answers the query in the first 100 words, original insight, scannable headings, no filler. Revise anything under 4. Stop when every score is 4+. Save the result as `draft`.
6. **Save as Markdown** — Finish: Save {{draft}} with a title tag and meta description to articles/{{keyword}}.md.

## Done when

Every step has run, each approval was given, and you've reported what was produced and where it was saved.

Values you type fill the chat prompt and the Open in links. The SKILL.md asks the agent for them at run time.