Outline Builder Skill for Codex

Outline Builder

outline/outline.refined.ok freezes a reviewed outline only while the marker is newer than the outline, all declared inputs, the defaults asset, and the generator. Upstream changes invalidate the marker and trigger a backed-up regeneration.

Build outline/outline.yml from either:

• outline/taxonomy.yml (legacy H3-first compatibility path)
• outline/chapter_skeleton.yml + outline/section_briefs.jsonl (section-first transition path)

Compatibility mode is active: this migration keeps the current output contract while moving intro/related defaults, Stage A bullet templates, and domain-specific comparison framing into references/ and assets/.

Load Order

Always read:

• references/overview.md
• references/stage_a_contract.md

Read by task:

• references/intro_related_patterns.md when changing Introduction / Related Work defaults
• references/examples_good.md and references/examples_bad.md for bullet calibration

Machine-readable asset:

• assets/outline_defaults.yaml

Inputs

Required:

• outline/taxonomy.yml

Optional human calibration only:

• ref/agent-surveys/STYLE_REPORT.md
• ref/agent-surveys/text/

Output

Keep the current output contract:

• outline/outline.yml

Compatibility mode

Current mode is reference-first with script compatibility:

• front-chapter defaults live in assets/outline_defaults.yaml
• Stage A bullet defaults and comparison-axis packs live in assets/outline_defaults.yaml
• examples and boundary rules live in references/
• scripts/run.py still owns outline materialization, section-first input selection, and placeholder-safe overwrite behavior

Script boundary

Use scripts/run.py only for:

• loading taxonomy or section-first structure inputs plus the defaults asset
• rendering the outline skeleton deterministically
• preserving existing non-placeholder outlines
• choosing comparison-axis packs from machine-readable defaults

Do not treat the script as the main place for:

• domain framing for Introduction / Related Work
• long bullet banks or writing exemplars
• prompt-heavy guidance about how a good outline should read

Output shape rules

Keep these stable:

• outline/outline.yml is a YAML list
• Introduction and Related Work remain the first two H2 sections
• each H3 subsection contains the Stage A bullets: Intent: / RQ: / Evidence needs: / Expected cites:
• each H3 subsection adds several topic-specific bullets after the Stage A fields
• the helper never overwrites non-placeholder user work

Quick Start

• uv run python .codex/skills/outline-builder/scripts/run.py --help
• uv run python .codex/skills/outline-builder/scripts/run.py --workspace <workspace>

Execution notes

When running this skill in compatibility mode, scripts/run.py currently reads:

• outline/taxonomy.yml, or
• outline/chapter_skeleton.yml + outline/section_briefs.jsonl
• assets/outline_defaults.yaml

The optional style references under ref/agent-surveys/ are for human calibration only:

• use ref/agent-surveys/STYLE_REPORT.md to sanity-check chapter counts / thickness
• skim ref/agent-surveys/text/ only to calibrate structure rather than wording

Script

Quick Start

• uv run python .codex/skills/outline-builder/scripts/run.py --workspace <workspace>

All Options

• --workspace <dir>
• --unit-id <id>
• --inputs <path1;path2>
• --outputs <path1;path2>
• --checkpoint <C*>

Examples

• uv run python .codex/skills/outline-builder/scripts/run.py --workspace <workspace>

Troubleshooting

• If Related Work still carries domain-specific framing, patch assets/outline_defaults.yaml before changing Python.
• If subsection bullets feel generic, review references/stage_a_contract.md and references/examples_good.md.
• If the outline is structurally valid but too fragmented, reroute to outline-budgeter rather than expanding this script.