Skip to main content
Turn your demand report into a content plan: one page per real demand cluster, each one the genuinely-best answer to a question people actually asked, citing the real Reddit threads that surfaced it. It’s formatted so AI search engines (ChatGPT, Perplexity, Google’s AI answers) cite you. content_plan is fully deterministic — no model, no embeddings, no network, no API key. It’s a pure projection of the report, so nothing is invented and no ranking is promised. 
from metalworks import Metalworks
mw = Metalworks()
research = mw.research("an affordable, jitter-free focus supplement for developers")

content = mw.content_plan(research)          # ContentPlan — deterministic, zero-key
for page in content.pages:
    print(page.target_phrase, page.page_kind)
    print("  ", page.stat_anchors)           # {'distinct_authors': N, 'mentions': M}

content.citation_strategy.reddit_targets     # the real Reddit permalinks to cite
content.citation_strategy.prompt_set         # example prompts to check citability against
From the CLI (report-id comes from metalworks research list): 
metalworks research content-plan <report-id>

​
What you give it / what you get back

You give it: a finished Research bundle (the report lives on .demand). That’s all — no keys, no model. You get back: a ContentPlan with one ContentPage per ranked demand cluster, plus a CitationStrategy. Every field is taken straight from the report:
  • target_phrase — the cluster’s own claim, normalized. Never a conjured keyword.
  • page_kind — a deterministic read of that phrase: comparison (vs/versus/or/best), guide (how/guide/tips), otherwise answer.
  • stat_anchors — the cluster’s real distinct-author and mention counts, so the page brief is as honest as the demand behind it.
  • faq — built verbatim from the report brief’s sub-questions, each with an empty answer slot for a citable answer to fill. The plan marks the slot; it never fabricates the answer.
  • outline — a fixed, answer-first section list: What people actually want → Common approaches → The honest answer → FAQ.
Reddit is one of the most heavily cited sources in AI search engines, so the legitimate play is to write the genuinely-best answer to a real question and cite the real threads that surfaced it. Two choices make a page citable:
  • Answer-first formatting plus a FAQPage JSON-LD stub, so an assistant can lift a clean, self-contained answer.
  • A CitationStrategy whose reddit_targets are the actual permalinks from your report’s top clusters — disclosed, real sources to cite, not placeholders. Its prompt_set holds example citability prompts derived from the target phrases.
Render the typed plan into shippable artifacts, both built mechanically from typed fields: 
from metalworks.research.marketing import render_content_markdown, render_faq_jsonld

md     = render_content_markdown(content)   # a markdown outline pack
jsonld = render_faq_jsonld(content)         # a schema.org FAQPage stub

​
When the result is thin

A content plan is only as deep as the report behind it: thin demand → few clusters → few pages. Because every value is projected from the report — no invented keywords, real stat anchors, real permalinks — the plan is exactly as defensible as the demand research. This is the legitimate way to get cited — the opposite of astroturf. Citing real permalinks and writing the best answer is authentic, disclosed work. Fake personas and invented backstories are excluded by design and prohibited by the usage policy. And note: this is a structural plan for citable content, not an SEO ranking guarantee.
Next: the authentic Reddit engagement loop — find threads worth joining and draft honest, disclosed replies. Or draft your launch assets from the same report. See also why you can trust the output.