Forge-core ExtractPrompt

Move bulk reference material from the main SKILL.md body into @-included companion files so the always-loaded content stays lean. USE WHEN an adopted or authored skill's body exceeds its information density because it inlines reference tables, pricing, catalogs, or per-variant guidance.

install

source · Clone the upstream repo

git clone https://github.com/N4M3Z/forge-core

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/N4M3Z/forge-core "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/ExtractPrompt" ~/.claude/skills/n4m3z-forge-core-extractprompt && rm -rf "$T"

manifest: skills/ExtractPrompt/SKILL.md

source content

ExtractPrompt

Move bulk reference material out of

SKILL.md

into companion files referenced with

includes. The SKILL.md stays the entrypoint an AI reads to decide behavior; companions load on demand when the AI needs the reference data. Referenced by ForgeAdopt as the

extract

transform.

Follows [ARCH-0002](docs/decisions/ARCH-0002 Skills Companion Files.md).

What to extract

Content shape	Extract to a companion?
Long reference table (model pricing, language-specific rules, catalogs)	Yes
Per-variant guidance (one section per framework, language, or ecosystem)	Yes, one companion per variant
Command reference (many commands, each with flags and examples)	Yes, if it spans more than a screen
Full example project	Yes, always
Workflow that decides HOW to use the skill	No, stays in `SKILL.md`
Constraints, red flags, anti-patterns	No, stays in `SKILL.md`
Decision tables with 3-7 rows	No, inline

Procedure

Identify candidate sections — content that is reference material rather than instruction.
For each candidate, move it to a companion file named by its scope (
```
PythonReference.md
```
,
```
Tables.md
```
,
```
ModelPricing.md
```
).

Replace the extracted block with an

include in

SKILL.md

## Language-specific guidance

@python-security.md
@typescript-security.md
@go-security.md

Keep the section heading in
```
SKILL.md
```
so the AI knows the companion exists.
Re-read
```
SKILL.md
```
alone. It should still be a complete instruction — the AI must be able to decide what to do without loading the companion.

Naming

Companions live in the same directory as
```
SKILL.md
```
PascalCase for conceptual companions (
```
TemplateReference.md
```
,
```
SchemaValidation.md
```
)
kebab-case for variant-keyed companions (
```
python-django-security.md
```
)
One companion per topic; do not bundle unrelated reference material

Constraints

```
SKILL.md
```
must remain complete as instruction on its own — companions are reference, not required loading
Never extract decision-making content; the AI needs it to choose which companion to read
Do not create a companion unless at least 200 words would move; smaller moves are just fragmentation
Keep companion files under 2000 words each; split further if they grow
Do not nest companions (a companion cannot
```
@
```
-include another companion) — compose at the SKILL.md level