iil-authoringfw 0.11.5

Authoring Framework — domain schemas and orchestration base for AI-assisted creative writing

iil-authoringfw — Authoring Framework

Domain schemas for AI-assisted creative writing applications.

Installation

pip install iil-authoringfw
# With aifw LLM integration:
pip install "iil-authoringfw[aifw]"
# All optional dependencies:
pip install "iil-authoringfw[all]"

Extras / Optional Dependencies

Extra	Dependencies	Purpose
aifw	iil-aifw>=0.11.4	LLM completion integration (quality_level/priority routing needs >=0.11.2)
yaml	pyyaml>=6.0	YAML template loading
promptfw	iil-promptfw>=0.5.5	Prompt template rendering
all	all above	Full feature set

Quick Start

from authoringfw import StyleProfile, CharacterProfile, WorldContext, get_format

# Style constraints for prompt injection
style = StyleProfile(tone="melancholic", pov="third_limited", tense="past")
constraints = style.to_constraints()

# Character context
alice = CharacterProfile(
    name="Alice",
    role="protagonist",
    personality_traits=["brave", "curious"],
    arc="From fear to courage",
)
print(alice.to_context_string())

# World context
world = WorldContext(
    title="The Shattered Realms",
    genre="fantasy",
    world_rules=["Magic costs life force", "Dragons are extinct"],
)
print(world.to_context_string())

# Format profiles (novel, essay, series, scientific)
roman = get_format("roman")
print(roman.style_constraints)

Schemas

• StyleProfile — tone, POV, tense, vocabulary, sentence rhythm
• CharacterProfile — name, role, traits, backstory, arc, relationships
• WorldContext — title, genre, setting, world rules, locations, lore
• VersionMetadata — immutable content snapshot with hash, semver, LLM metadata
• PhaseSnapshot — project state at a workflow phase boundary

Format Profiles

Built-in formats: roman (alias novel), essay, serie, scientific, nonfiction, academic, screenplay, short_story, blog_post, podcast_script. The registry (authoringfw.formats.base.FORMAT_REGISTRY) is the source of truth.

from authoringfw.formats.base import get_format, WorkflowPhase

novel = get_format("roman")
outline_steps = novel.steps_for_phase(WorkflowPhase.OUTLINE)

Adapter Interfaces

Protocol-based adapters — no inheritance required:

from authoringfw.adapters.interfaces import IStyleAdapter

class MyStyleAdapter:
    async def get_profile(self, style_id): ...
    async def analyze_text(self, text): ...
    def generate_style_constraints(self, profile): ...
    async def score_conformity(self, text, profile): ...

adapter = MyStyleAdapter()
assert isinstance(adapter, IStyleAdapter)  # True via @runtime_checkable

Changelog

See CHANGELOG.md.