agentleFS
Vumb-VibeCoderVumb-VibeCoder/hook-everything/skills/writing-hooks/SKILL.md

writing-hooks

Skill19 starsChanged 9 days agoMITView on GitHub

Craft invisible openers, False-Summit retention bridges, 90/10 Socratic generation gaps, and Peak-End closers in any language. Sub-skill: load only when routed from the hook-everything entry skill, not for general writing or editing.

2 other projects use a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 2 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Use this guide to engineer continuous, invisible pull from the first word to the final period.

    line 9
  2. Very specific:

    Never start with a macro-generalization ("Since the dawn of...", "In modern software development...").

    line 15
  3. Distinctive:

    Instead of stating your thesis and then listing 3 supporting bullets (which puts the reader into passive skimming mode), invert the order:

    line 43
  4. Very specific:

    Ban Textbook Subheads: Never use one-word labels (Background, Implementation, Results, Conclusion).

    line 47
  5. Very specific:

    Hook Skimmers in Captions (M27): Write every code-block comment, table header, and image caption as a self-contained insight so a skimmer …

    line 48
  6. Very specific:

    Never end with a summary paragraph that starts with "In conclusion" or repeats what the reader just read.

    line 64

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Writing Hooks & Long-Form Retention Architecture

Most writing advice focuses only on the first sentence. In practice, reader drop-off often spikes around Paragraph 3 (working heuristic, verify with your own data) when the opener's tension evaporates into background filler, and at the first H2 heading when the article turns into a dry textbook list.

Use this guide to engineer continuous, invisible pull from the first word to the final period.


Phase 1: Openers That Bypass the Reader's Shield

Never start with a macro-generalization ("Since the dawn of...", "In modern software development..."). Select an opener chord (skills/hook-compounding) that pairs immediate tension with a Trust Damper:

  1. In-Medias-Res + Pratfall (M2 + M20) drops into a specific moment where a capable narrator hit an unexpected wall:
    • "I deleted the production database at 4:55 PM on a Friday. Nobody noticed until Monday morning, which was somehow worse."
  2. Asymmetric Anomaly + Boundary (M4 + M15) pairs two verifiable facts that violate expectations, and states who cares:
    • "Our slowest API endpoint had zero database queries, twelve lines of Python, and cost us $4,100 a month in idle worker timeouts."
  3. Inoculated Counter-Thesis (M21 + M3) validates the reader's current default belief, then shows the exact measurement where it breaks:
    • "Everyone tells you to index foreign keys first. On tables with fewer than 50 distinct values, that index makes Postgres 3x slower."

Phase 2: Surviving Paragraph 3 (The "No-Backstory" Bridge)

Writers often ruin a strong opener by retreating into 300 words of history ("Before I explain how we fixed the database, let's review the history of our company's architecture..."). That kills momentum immediately.

  • The 150-Word First Payoff Rule delivers one concrete, usable insight or forensic clue within the first 150 words before giving any background context.
  • Flashback by Necessity Only introduces background details only at the exact sentence where the reader needs that detail to understand the next action.

Phase 3: Mid-Piece Retention Mechanics (Keeping Readers Through 2,000+ Words)

1. The "False Summit" Turn (Mid-Article Re-Hook)

Around the 40% mark of an article or case study, readers think they have figured out the whole answer and prepare to close the tab.

  • Execution Solve the initial problem cleanly at the 40% mark, then immediately reveal the Second-Order Consequence that the obvious fix created.
  • Example - "Adding the local LRU cache dropped median latency from 310ms to 12ms. We celebrated on Thursday. On Friday morning, our memory usage hit 98% because every user's session object held a reference to the entire tenant graph."

2. The 90/10 Socratic Generation Gap (Slamecka & Graf 1978)

Instead of stating your thesis and then listing 3 supporting bullets (which puts the reader into passive skimming mode), invert the order:

  • Lay out Clue 1 and Clue 2 side-by-side so the reader's brain connects the dot half a second before your next sentence confirms it (M11). When readers generate the conclusion themselves, they experience a rush of personal insight and stay locked in.

3. Assertion-Tension Subheads & Caption Hooks

  • Ban Textbook Subheads Never use one-word labels (Background, Implementation, Results, Conclusion). Write every H2/H3 as a declarative assertion or an unresolved tension (Why the 30-second timeout made the outage twice as long).
  • Hook Skimmers in Captions (M27) Write every code-block comment, table header, and image caption as a self-contained insight so a skimmer who scrolls down to look at a diagram gets pulled back into the prose.

Phase 4: Pacing Architecture by Length

Piece LengthHook & Payoff RhythmStructural Rule
Short-Form (< 300 words) (Emails, Social Posts, Release Notes)Single Chord: 1 sharp opener (< 20 words) -> 1 concrete mechanism/example -> 1 immediate takeaway or micro-action.Zero section headings, zero throat-clearing, zero summary closer.
Medium-Form (300 - 1,200 words) (Blog Posts, Case Studies, Memos)3-Chord Arc: Opener Chord -> First Payoff at word 150 -> False-Summit Turn at word 500 -> Peak-End Closer.Include at least 1 visual eye-brake (table, diff, or numbered forensic trace) around the midpoint.
Long-Form (1,200 - 4,000+ words) (Deep Dives, Whitepapers, Guides)Rotated Chord Sequence: Rotate Chord 3 -> Chord 5 -> Chord 1 -> Chord 4 across major sections. Close every open loop within 250 words before opening the next.Audit in 300-word blocks (anti-slop Rule 71) so the prose never degrades into repetitive LLM filler in the second half.

Phase 5: Peak-End Closers (Fredrickson & Kahneman 1993)

Never end with a summary paragraph that starts with "In conclusion" or repeats what the reader just read. End at the highest point of utility:

  • The 60-Second Diagnostic Closer gives the reader one exact command, search query, or mental check they can run in the next 60 seconds to see if they have the same issue.
  • The Resonant Artifact Closer ends on a concrete physical detail that encapsulates the lesson ("I still keep a yellow sticky note taped to my monitor bezel: check the hostname before you hit enter.").

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.