Menu

Browse all skills

Content agent skill

Content Brief

produce a decision document that the `content` or `pillar-article` skill can execute with no further guessing. A brief is not a summary of the topic. It answers every open question about what the page should be: who it is for, what it must cover, what it must prove, what it should look like, and what it links to.

View raw SKILL.mdView package on GitHub

Shared rules:

When to use this skill

Use when the user wants a brief, outline, or plan for a page before writing, or when a writer or another skill needs one.

Do not use when:

  • The user wants the finished article now. Use content, which can build its own outline.
  • The topic needs a hub-and-spoke plan. Use pillar-article, which includes spoke planning.
  • The user wants a full multi-asset industry package. Use industry-content-production.

Required inputs

target_query: ""
brand:
  name: ""
  website: ""
  offer: ""
market:
  country: ""
  language: ""
voice_sample: ""           # optional: URL or text of existing brand copy
existing_pages: []         # optional
prior_findings: ""         # optional: competitor-audit / keyword-research output
first_hand_assets: []      # optional: data, case studies, experts the brand can contribute

Workflow

- [ ] 1. Query type, funnel stage, scope (is / is not) stated
- [ ] 2. Live results + prior findings reviewed; whitespace angle chosen
- [ ] 3. Audience, prior knowledge, and reader outcome stated
- [ ] 4. Voice reference identified
- [ ] 5. Outline with formats, depth, and evidence required per section
- [ ] 6. FAQ, internal links, title options, and CTA planned
- [ ] 7. Brief assembled; open questions listed
  1. Classify.
    • Assign the query type and funnel stage using references/query-types.md.
    • State in one sentence what the page is, and in one sentence what it is not. Scope creep in a brief produces an unfocused draft.
  2. Research.
    • Review live results for the exact query, or reuse competitor-audit / keyword-research findings, with retrieval dates.
    • Record which format is winning, what the strongest pages do structurally, and which sources AI engines cite for this topic, if known.
    • Choose a whitespace angle the brand can credibly own, preferably based on first-hand or non-commodity material. If you have no live access, label the research as assumption.
  3. Audience and intent.
    • Describe the actual searcher in concrete terms: what they already know and what they are missing.
    • State in one sentence what the reader should be able to do or decide after reading. If that takes more than one sentence, split the topic.
  4. Voice. Point to a specific brand voice sample provided by the user, or to existing on-site copy by URL. If neither exists, list that as an open question. Do not invent a generic tone label.
  5. Outline.
    • Planned H1, matching the query near verbatim.
    • Planned H2s written as real sub-questions, each with:
      • a format (table, steps, list, grouped by period, prose)
      • a depth guide (approximate words)
      • the evidence required, such as "current price range from 2+ provider pages" or "first-hand quote from brand expert"
    • Evidence requirements prevent writers from filling gaps with invention.
  6. Supporting plan.
    • 3–6 FAQ questions drawn from adjacent long-tail variants the body will not answer.
    • Internal links out, and links back from a pillar page if one exists.
    • 2–3 title tag options following the pattern for the query type.
    • One CTA that matches the funnel stage.
  7. Assemble using the template. List open questions the user must answer before drafting.

Output template

# Brief: <target query>

| Field | Value |
|---|---|
| Query type / funnel stage | |
| Scope — is | |
| Scope — is not | |
| Audience & prior knowledge | |
| Reader outcome (one sentence) | |
| Angle / whitespace | |
| Voice reference | |
| CTA | |

## Competitive context (accessed <date>)
| URL | Format | Strength | Gap |

## Outline
| # | H2 (sub-question) | Format | ~Words | Evidence required | Source hints |

## FAQ
## Internal links (out / in)
## Title options
## Sources
| Claim/area | URL | Accessed | Confidence |
## Open questions for the user

Approval gates

Get user confirmation of scope, angle and CTA before handing the brief to a writing skill when the topic is commercially sensitive. That includes pricing, competitor comparisons, and regulated topics.

Stop conditions

  • The target query has mixed or unclear intent that research cannot resolve. Present the options instead.
  • The brand cannot credibly support the only viable angle, for example because it has no product or expertise in the topic.

Anti-patterns

  • Summarizing the topic instead of making the decisions.
  • Briefing from assumption without reviewing live results, or without labeling the gap.
  • Leaving formats or evidence requirements unassigned per section.
  • Generic personas that could apply to any query.
  • Covering two topics in one brief.
  • Planning FAQPage or HowTo schema for rich-result gains. Those rich results are gone.

Completion checklist

  • Query type, funnel stage and is/is-not scope stated
  • Research dated, or labeled assumption. Angle justified
  • Concrete audience and a one-sentence reader outcome
  • Specific voice reference, or an open question
  • Every H2 has a format, depth and evidence requirement
  • FAQ, internal links, title options and CTA planned
  • Open questions listed