outline-builder_skill

This skill converts a taxonomy into a bullets-only, checkable outline with sections and subsections, ensuring each subsection has at least three actionable
  • Python

109

GitHub Stars

1

Bundled Files

2 months ago

Catalog Refreshed

4 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill willoscar/research-units-pipeline-skills --skill outline-builder

  • SKILL.md7.0 KB

Overview

This skill converts a taxonomy (outline/taxonomy.yml) into a bullets-only, checkable outline (outline/outline.yml) suitable for mapping evidence and driving writing. It produces paper-like sections and subsections where every subsection contains verifiable Stage A bullets (Intent, RQ, Evidence needs, Expected cites) plus concrete, topic-specific coverage items. The output enforces guardrails: bullets-only, no TODOs, and at least three non-template bullets per subsection.

How this skill works

The tool reads outline/taxonomy.yml, translates nodes into H2/H3 headings, and emits a bullets-only outline with Section and Subsection bullets. For each H3 it writes the Stage A contract bullets and 2–6 subsection-specific bullets that encode comparisons, evaluation anchors, failure modes, and scope boundaries. It also supports optional style calibration from ref/agent-surveys to keep section counts paper-like and appends global Discussion/Conclusion later in the pipeline.

When to use it

  • You have a taxonomy and need a mappable chapter/subsection skeleton (no prose).
  • You want each subsection to state explicit coverage requirements and evidence needs.
  • Preparing to assign papers or experiments to outline axes for evidence-first drafting.
  • Replacing generic scaffold bullets with topic-specific, checkable items.
  • Avoid running when an approved, mappable outline already exists.

Best practices

  • Keep final ToC compact (target ~5–6 H2 including Intro/Related) to remain paper-like.
  • Ensure every H3 includes Stage A bullets: Intent, RQ, Evidence needs, Expected cites.
  • Replace template language with domain terms, datasets, benchmarks, and failure modes.
  • Enforce scope cues in bullets (in-scope vs out-of-scope) to prevent silent drift.
  • Remove all TODO/TBD/FIXME markers before marking outline refined.

Example use cases

  • Generate an evidence-mapping outline from a taxonomy for a survey paper.
  • Create a writer-friendly subsection skeleton that a downstream author can draft to without guessing.
  • Prepare a mapping sheet where each bullet corresponds to a paper or experiment to collect.
  • Run a check to find missing Stage A fields or placeholder bullets before research synthesis.
  • Refine a baseline scaffold into a vetted outline and signal completion with outline.refined.ok.

FAQ

Merge adjacent H3s into thicker subsections; the architect role budgets H2/H3 counts to avoid explosion.

How many bullets per subsection are required?

Every subsection must include the four Stage A bullets plus at least three additional topic-specific, checkable bullets.

Built by
VeilStrat
AI signals for GTM teams
© 2026 VeilStrat. All rights reserved.All systems operational
outline-builder skill by willoscar/research-units-pipeline-skills | VeilStrat