Agent Skills: Outline Budgeter (NO PROSE)

|

UncategorizedID: willoscar/research-units-pipeline-skills/outline-budgeter

Install this agent skill to your local

pnpm dlx add-skill https://github.com/WILLOSCAR/research-units-pipeline-skills/tree/HEAD/.codex/skills/outline-budgeter

Skill Files

Browse the full folder contents for outline-budgeter.

Download Skill

Loading file tree…

.codex/skills/outline-budgeter/SKILL.md

Skill Metadata

Name
outline-budgeter
Description
|

Outline Budgeter (NO PROSE)

Purpose: prevent the most common survey failure mode: H3 explosion (too many tiny subsections) leading to a thin, outline-like PDF.

This skill rewrites outline/outline.yml into a paper-like budget:

  • Final ToC target: ~6–8 H2 sections (Intro / Related Work + 3–4 core chapters + Discussion + Conclusion)
  • H3 target: fewer, thicker subsections (per draft_profile: survey<=10, deep<=12)

Important: Discussion/Conclusion are appended in C5 merge (global sections), so the outline itself should usually be <=6 H2.

Inputs

  • outline/outline.yml
  • Optional (helps make merges evidence-aware):
    • queries.md (optional: if it sets draft_profile, use it to choose the H3 budget)
    • outline/mapping.tsv
    • outline/coverage_report.md
    • GOAL.md

Outputs

  • outline/outline.yml (updated in place)
  • outline/OUTLINE_BUDGET_REPORT.md (bullets-only; what was merged and why)

Workflow (NO PROSE)

  1. Read the outline and compute a simple budget snapshot:
  • If queries.md sets draft_profile (survey/deep), use it to decide the H3 budget target.
  • H2 count (excluding Discussion/Conclusion, which are not in the outline)
  • Total H3 count
  • H3 count per H2 chapter
  1. Decide a merge plan (structure-first, evidence-aware):
  • Prefer merging adjacent H3s that share similar axes/keywords.
  • Prefer merging H3s with weak mapping coverage (if mapping.tsv exists).
  • If outline/coverage_report.md exists (from outline-refiner), use it to identify weak-coverage or high-reuse subsections to merge.
  • Use GOAL.md as the scope constraint: avoid merges that mix distinct research questions or scope boundaries.
  • Prefer moving fine-grained distinctions into bullets/axes instead of creating new subsections.
  1. Apply merges in outline/outline.yml:
  • Merge titles into a clearer, thicker subsection title.
  • Merge bullets (dedupe templates; keep Stage A fields: Intent/RQ/Evidence needs/Expected cites).
  • Keep ids stable when possible.
    • If you must change ids, record it explicitly in the report and assume mapping.tsv must be regenerated.
  1. Write outline/OUTLINE_BUDGET_REPORT.md:
  • Before/after counts.
  • List of merges (old ids/titles -> new id/title).
  • Any risks (e.g., mapping reset required).

Quality checklist

  • [ ] No placeholders (TODO//(placeholder)).
  • [ ] Outline budget matches the paper-like target.
  • [ ] Each remaining H3 is thick enough to sustain evidence-first writing (its bullets mention concrete comparisons + eval anchors + failure modes).

Troubleshooting

Issue: merging makes a subsection too broad

Fix:

  • Keep one H3, but split its bullets into explicit comparison axes and required evidence fields; defer fine-grained splits to later if evidence is strong enough.

Issue: mapping breaks after id changes

Fix:

  • Rerun section-mapper to regenerate outline/mapping.tsv, then rerun outline-refiner.