Bookends — Strong Openings & Closings

The Bookends module (fabric_workflow_sdk.bookends) provides prompt-building utilities that generate rich, structured guidance for powerful content openings and closings. It works across all artifact types — videos, images, captions, blog posts, series, and ad sequences.

Quick Start

from fabric_workflow_sdk import build_bookends_prompt

# Get both opening and ending guidance in one call
prompt_block = build_bookends_prompt(
    artifact_type="video",
    platform="tiktok",
    mood="edgy",
    hook_type="question",
)

# Inject into your LLM prompt
full_prompt = f"""Write a script about productivity hacks.

{prompt_block}
"""

Core Concepts

Artifact Types

The module supports these artifact types, each with tailored guidance:

Artifact Type	Opening Focus	Closing Focus
`video`	First 3 seconds — life or death for retention	Direct-to-camera CTA with hook coherence
`image`	2-4 word overlay text, emotional not descriptive	Caption CTA matching visual register
`blog`	Hook as opening line, not topic sentence	Closing paragraph earns the CTA
`caption`	First line above the fold stops the scroll	Single clear action, tone-matched
`series`	Part 1: strongest hook. Part 2+: callback	Specific cliffhanger (not “stay tuned”)
`ad`	Phase-aware: pain recognition first	Escalating: no CTA → soft → hard

CTA Types

12 closing strategies that work across all artifact types:

Type	Strategy	Example
`follow`	Subscription ask	”Follow — I drop one of these every day.”
`save`	Bookmark urgency	”Save this before the algorithm buries it.”
`comment`	Engagement prompt	”Drop your hottest take below.”
`share`	Social spread	”Tag someone who needs to hear this.”
`challenge`	Action dare	”Try this for 7 days. Come back and tell me.”
`cliffhanger`	Serial hook	”But here’s the part nobody talks about… tomorrow.”
`question`	Identity mirror	”So which one are you?”
`identity`	Tribal belonging	”If you made it this far, you’re different.”
`proof_request`	Skeptic redirect	”Don’t believe me? Try it yourself.”
`value_promise`	Content contract	”I break down one of these every week.”
`mic_drop`	No ask, devastating final line	Stands alone — no CTA needed.
`callback`	Loop to opening	References the hook, closes the circle.

Ending Styles

8 arc patterns that describe how to transition into the close:

Style	Arc
`mic_drop`	Build tension, release with one devastating line
`callback_loop`	Reference the opening, circular “oh wow” moment
`cliffhanger_bridge`	Resolve this piece, plant seed for next
`emotional_peak`	Escalate emotion, CTA rides the wave at peak
`pattern_break`	Shift tone/pace abruptly to snap attention
`social_proof_close`	End with evidence, CTA while credibility is highest
`story_resolve`	Complete narrative arc, CTA as natural next chapter
`rapid_recap`	Quick-fire summary, CTA as “now act” punchline

Opening Styles

8 opening frameworks:

Style	Approach
`bold_claim`	Strong, specific, slightly uncomfortable statement
`question`	Ask something the audience cannot ignore
`contrarian`	Challenge conventional wisdom head-on
`story_drop`	Start mid-story, no preamble
`pain_point`	Name their exact frustration
`shock_stat`	Lead with a surprising number or fact
`identity_call`	”If you’re a [type]…” — instant self-selection
`pattern_interrupt`	Break the expected format

API Reference

`build_bookends_prompt()`

The main entry point. Returns both opening and ending prompt blocks combined.

from fabric_workflow_sdk import build_bookends_prompt

block = build_bookends_prompt(
    artifact_type="video",     # video | image | blog | caption | series | ad
    platform="tiktok",         # tiktok | youtube shorts | instagram | linkedin | twitter | threads
    mood="edgy",               # mood/tone descriptor
    hook_type="question",      # optional — drives CTA strategy recommendation
    emotional_trigger="curiosity",  # optional — refines CTA selection
    cta_style=None,            # optional — formula override (free text)
    ending_style=None,         # optional — force a specific ending arc
    opening_style=None,        # optional — force a specific opening framework
)

`build_opening_prompt()`

Returns only the opening prompt block.

from fabric_workflow_sdk import build_opening_prompt

opening = build_opening_prompt(
    artifact_type="caption",
    platform="instagram",
    mood="conversational",
    hook_type="pain_point",
    emotional_trigger="empathy",
    opening_style="story_drop",  # optional override
)

`build_ending_prompt()`

Returns only the ending prompt block.

from fabric_workflow_sdk import build_ending_prompt

ending = build_ending_prompt(
    artifact_type="video",
    platform="tiktok",
    mood="edgy",
    hook_type="contrarian",
    emotional_trigger="outrage",
    cta_style=None,             # optional formula override
    ending_style="mic_drop",    # optional override
)

`recommend_cta_strategy()`

Pure mapping logic — recommends the best CTA strategy based on hook type and emotional context.

from fabric_workflow_sdk import recommend_cta_strategy

strategy = recommend_cta_strategy(
    hook_type="question",
    emotional_trigger="curiosity",
    artifact_type="video",
    platform="tiktok",
)
# Returns:
# {
#     "cta_type": "callback",
#     "ending_style": "callback_loop",
#     "reasoning": "A question hook creates a natural bridge to a callback CTA...",
#     "examples": ["Remember that question I asked at the start? Now you know the answer.", ...]
# }

Hook-to-CTA Priority Matrix

The recommendation engine maps hook types to optimal CTA strategies:

Hook Type	Recommended CTA	Recommended Ending Style
`question`	`callback`	`callback_loop`
`contrarian`	`mic_drop`	`mic_drop`
`cliffhanger`	`cliffhanger`	`cliffhanger_bridge`
`pain_point`	`challenge`	`emotional_peak`
`shock`	`proof_request`	`pattern_break`
`social_proof`	`share`	`social_proof_close`
`story`	`callback`	`story_resolve`
`fomo`	`save`	`rapid_recap`

Platform adjustments are applied automatically (e.g., Instagram favors save over follow).

Usage Patterns

Video Scripts

The ai_shorts pipeline uses bookends automatically. Pass optional overrides via input:

input = {
    "topic": "Why cold showers boost productivity",
    "hook_type": "contrarian",         # drives CTA strategy
    "emotional_trigger": "surprise",   # refines recommendation
    "cta_style": "mic_drop",           # force a specific CTA type
}

Formula-Driven Videos

When using formula_shorts, the formula’s cta_style field flows through to build_ending_prompt() automatically:

# In your formula JSON:
{
    "script": {
        "hook_type": "question",
        "cta_style": "always end with a challenge — dare the viewer to try it"
    }
}

Blog Posts & Captions

Use build_bookends_prompt() in your content generation prompts:

from fabric_workflow_sdk import build_bookends_prompt

bookends = build_bookends_prompt(
    artifact_type="blog",
    platform="linkedin",
    mood="professional",
    hook_type="bold_claim",
)

prompt = f"""Write a LinkedIn article about remote work productivity.

{bookends}

Write 800-1200 words. Include specific examples and data.
"""

Series Content

For multi-part series, the bookends module provides cliffhanger/callback guidance:

from fabric_workflow_sdk import build_ending_prompt, build_opening_prompt

# Part 1: Strong hook, cliffhanger ending
part1_ending = build_ending_prompt(
    artifact_type="series",
    platform="twitter",
    mood="dramatic",
    ending_style="cliffhanger_bridge",
)

# Part 2+: Callback opening, builds on previous
part2_opening = build_opening_prompt(
    artifact_type="series",
    platform="twitter",
    mood="dramatic",
    opening_style="story_drop",
)

Ad Sequences

The module is phase-aware for advertising content:

from fabric_workflow_sdk import build_ending_prompt

# Phase 1: Awareness — no CTA
phase1 = build_ending_prompt("ad", "instagram", "empathetic")

# Phase 3: Conversion — hard CTA with friction removal
phase3 = build_ending_prompt("ad", "instagram", "urgent", cta_style="direct conversion")

Platform Intelligence

The module includes platform-specific CTA guidance that reflects how each platform’s algorithm rewards different engagement signals:

TikTok

Comments + shares drive distribution. Duet/stitch CTAs work. The algorithm treats comments as intent signals.

YouTube Shorts

Subscribe + like signals matter most. “Watch the full video” bridges to long-form. End screens don’t work on Shorts.

Instagram

Saves are the strongest signal. Story shares amplify reach. Profile visits from Reels drive follower growth.

Thoughtful comments signal quality. The audience punishes engagement-bait. Invite professional perspective, not agreement.

Constants

All taxonomies are exported as lists for use in extraction, classification, or UI:

from fabric_workflow_sdk import CTA_TYPES, ENDING_STYLES, OPENING_STYLES

print(CTA_TYPES)
# ['follow', 'save', 'comment', 'share', 'challenge', 'cliffhanger',
#  'question', 'identity', 'proof_request', 'value_promise', 'mic_drop', 'callback']

print(ENDING_STYLES)
# ['mic_drop', 'callback_loop', 'cliffhanger_bridge', 'emotional_peak',
#  'pattern_break', 'social_proof_close', 'story_resolve', 'rapid_recap']

print(OPENING_STYLES)
# ['bold_claim', 'question', 'contrarian', 'story_drop', 'pain_point',
#  'shock_stat', 'identity_call', 'pattern_interrupt']

Defaults are available via constants:

from fabric_workflow_sdk import DEFAULT_CTA_TYPE, DEFAULT_ENDING_STYLE, DEFAULT_OPENING_STYLE

# DEFAULT_CTA_TYPE = "follow"
# DEFAULT_ENDING_STYLE = "emotional_peak"
# DEFAULT_OPENING_STYLE = "bold_claim"