Research Output Formats

Research output formats

Use When

• Select a research deliverable family and its academic or non-academic variant after the audience and decision are known.

Do Not Use When

• Do not use for method selection or to override a mandatory commissioning template.

Inputs

Workflow

1. Identify the reader action and check for a mandated form.
2. Select one format and variant; stop if governing requirements conflict.
3. Load that reference, map sections, draft from verified findings, and test compliance.
4. If rendering or citation checks fail, recover through correction and rerun the checks.

Outputs

Evidence Produced

Capability Contract

Selection and review default to read-only. Editing/rendering requires authority; submission, publication, legal sign-off, or representation as counsel/expert requires explicit authority and qualification.

Degraded Mode

Without the mandate, style guide, sources, or renderer, provide a defensible section map and mark unassessed checks. Do not certify compliance.

Decision Rules

Quality Standards

The selected form is explicit, source-traceable, and tested against its governing audience or template.

Anti-Patterns

• Choosing by title alone. Fix: route from audience and decision.
• Mixing two structures. Fix: select one mandate.
• Ignoring a supplied template. Fix: map it first.
• Drafting before citation selection. Fix: choose the variant first.
• Certifying without authority. Fix: state the review boundary.

Worked Example

A ministerial decision request routes to a policy brief rather than an academic paper; map recommendation and evidence sections and reserve technical depth for annexes.

References

• Academic formats
• Advocacy and public formats
• Commercial formats
• Analytical and professional formats
• Academic and non-academic variants

The engine produces many document types. They share evidence discipline but differ in purpose, audience, structure, voice, and citation regime. Picking the wrong format is the most common reason a technically correct piece of research lands badly.

This skill is a router. Use it to:

1. Identify the format the deliverable belongs to.
2. Decide whether the academic or non-academic variant is appropriate.
3. Load the matching reference for structure, length, voice, and citation rules.

It does not replace academic-writing, business-writing, report-and-proposal-craft, analytic-tradecraft, consulting-delivery, or online-legal-research — it sits above them and tells you which to invoke for the given output.

The non-negotiable guardrail

Every format defers to source-evaluation/references/evidence-discipline.md. No format excuses fabrication, even popular ones (op-eds, pamphlets, product descriptions). Format changes the voice, never the truth standard.

The two axes

Every output sits on two axes:

Resolve both before drafting a single line.

Selection router

Match the deliverable to a row, then load the named reference.

After format is fixed, always load references/academic-vs-nonacademic-variants.md to choose voice and citation regime.

Reference router

references/
  academic-formats.md                   essays, papers, theses, dissertations,
                                        research proposals, literature reviews
  advocacy-and-public-formats.md        pamphlets, op-eds, policy briefs
  commercial-formats.md                 white papers, market research reports,
                                        market analyses, product descriptions,
                                        case studies, feasibility studies
  analytical-and-professional-formats.md intelligence reports, legal opinions
  academic-vs-nonacademic-variants.md   the variant axis applied to every format

Load only the family file relevant to the deliverable, plus the variants reference. Loading every file is wasteful and dilutes attention.

When to invoke this skill

• User asks for "a paper", "a white paper", "a brief", "a report", "an opinion", "a study" — these words map to different formats and you must disambiguate.
• The same research has to be repackaged for a new audience (e.g., an academic paper turned into a policy brief).
• Output is being generated by a sub-agent and you're writing the brief — name the format and variant in the brief.
• Reviewer says "this reads like an essay, not a brief" — you have a format mismatch, not a writing problem.

When NOT to invoke

• Format is obvious and already locked (e.g., the user gave you a journal template, a publisher's house style, or a regulator's prescribed form). Use that template directly.
• The work is internal, throwaway, or conversational. No format selection needed.

Decision flow

deliverable requested
        │
        ▼
  Is the format already prescribed (template, house style, statutory form)?
        │
   yes ─┴─ use that template, ignore this skill
        │
   no   ▼
  Match deliverable signal to selection-router row
        │
        ▼
  Load matching family reference for structure
        │
        ▼
  Load academic-vs-nonacademic-variants.md
        │
        ▼
  Pick variant; lock voice, citation regime, length
        │
        ▼
  Compose deliverable; cite-check against source-evaluation

Common mistakes

See also

• report-and-proposal-craft — long-form structure and proposal-specific patterns
• academic-writing — voice, citation, and argument standards for the academic variants
• business-writing — voice and structure for the non-academic variants
• analytic-tradecraft — drives the intelligence-report family
• online-legal-research — drives the legal-opinion family
• consulting-delivery — drives the market-analysis and feasibility-study families
• professional-word-output / python-document-generation — terminal renderers for any of these formats