agentleFS
vybzaivybzai/spitfire-storybook/skills/brutally-honest-book-builder/SKILL.md

brutally-honest-book-builder

Skill0 starsChanged 5 months agoMITView on GitHub

Turn a user's own connected data and assistant memory into a book-length, deeply personalized, brutally honest self-development guide. Use this when the user says things like "build me a personalized book", "write a brutally honest self-development book from my data", "turn my data into a book", "make me a development guide based on how I actually work", or asks for an evidence-based personal-growth book grounded in their real email, chat, tasks, calendar, meeting notes, and memory. Runs a consent-gated, read-only pipeline: Collect -> Diagnose -> Generate (~30-40 chapters, ~120k+ words) -> Build (PDF + EPUB). The private diagnostic profile never ships in the compiled book.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Get explicit consent before reading anything: Before touching any connected source, tell the user exactly which sources you intend to read …

    line 51
  2. Distinctive:

    Read-only: Never send, post, modify, delete, or create anything in the user's connected sources.

    line 56
  3. Distinctive:

    NOT copy raw private content (full emails, message logs, names of third parties, financial figures, client identities) into any file, and …

    line 59
  4. Distinctive:

    Let the user stop or redact at any point: If they want a source dropped or a pattern removed, comply immediately.

    line 67
  5. Very specific:

    Before starting, create a .gitignore in the book working directory (if not already present) that excludes the private profile and any …

    line 70
  6. Distinctive:

    Keep it git-ignored and out of the build.

    line 100

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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

Brutally Honest Book Builder

This skill helps you turn a user's OWN data into a long-form, deeply personalized self-development book. The book is written in the voice of a no-bullshit mentor: harsh and direct on the surface, genuinely on the user's side underneath. The specificity comes from evidence in the user's real working life, not generic advice.

The pipeline has four steps. Follow them in order:

  1. Collect — read across the user's connected sources to understand how they actually operate.
  2. Diagnose — synthesize a private, evidence-based profile of strengths and shortcomings.
  3. Generate — run a multi-agent workflow to write ~30-40 chapters from a strict shared template seeded with the profile.
  4. Build — compile the Markdown into PDF and EPUB with pandoc.

Detailed instructions for each step live in the reference/ files. Load each one when you reach that step — do not read them all up front.

  • Step 1: reference/01-collect-data.md
  • Step 2: reference/02-build-profile.md
  • Step 3: reference/03-generate-book.md
  • Step 4: reference/04-build-pdf-epub.md

Build assets live in assets/: metadata.yml (title/author/PDF + EPUB settings), Makefile (compile targets), and epub.css (EPUB styling).


Consent & Privacy (read this first — non-negotiable)

This skill reads the user's private life. Treat that as a responsibility, not a convenience.

  • Get explicit consent before reading anything. Before touching any connected source, tell the user exactly which sources you intend to read (e.g. email, team chat, task/issue tracker, calendar, meeting transcripts, notes/wiki, long-term memory) and wait for a clear yes. If a source is not approved, do not read it.
  • Read-only. Never send, post, modify, delete, or create anything in the user's connected sources. You are observing, not acting.
  • Summarize, never exfiltrate. Capture patterns and paraphrased evidence. Do NOT copy raw private content (full emails, message logs, names of third parties, financial figures, client identities) into any file, and never into the book.
  • The profile is private and stays private. Step 2 produces a working profile file. It is git-ignored, lives outside the book sources, and is NEVER included in the compiled PDF/EPUB. Verify this during the build (Step 4).
  • The user opts into harshness. The book is blunt by design. Confirm the user wants that tone before generating. Harsh means honest and constructive, never cruel or contemptuous.
  • Let the user stop or redact at any point. If they want a source dropped or a pattern removed, comply immediately.

Before starting, create a .gitignore in the book working directory (if not already present) that excludes the private profile and any intermediate notes, for example:

profile.md
profile/
notes/
*.private.md

Step 1 — Collect

Read reference/01-collect-data.md.

With consent secured, survey the approved sources to understand how the user really works: how they communicate, where they follow through and where they drop things, how they spend time, recurring frictions, and recurring strengths. Produce concise, paraphrased pattern notes — not raw dumps.

Step 2 — Diagnose

Read reference/02-build-profile.md.

Synthesize the collected patterns into a single private profile.md: top strengths, top critical shortcomings, recurring behavioral patterns, each backed by concrete but paraphrased evidence. This profile is the engine of the whole book — its specificity is what makes the book feel written for this one person. Keep it git-ignored and out of the build.

Step 3 — Generate

Read reference/03-generate-book.md.

Plan ~30-40 chapters that map directly onto the profile's findings. Then use a multi-agent workflow — one writer agent per chapter, all working from a single strict chapter template seeded with the relevant profile excerpts — to write each chapter at ~3,000-4,500 words (~120k+ words total). Every chapter follows the same structure: cold open, hard truth, teaching (with Mermaid diagrams and tables), worked example, real case study, a "Harsh Mirror" self-assessment with a 1-5 scorecard, drills, common mistakes, key takeaways, and a weekly commitment.

Step 4 — Build

Read reference/04-build-pdf-epub.md.

Compile the Markdown chapters into PDF (via pandoc + xelatex) and EPUB, using the files in assets/ (metadata.yml, Makefile, epub.css). Optionally render Mermaid diagrams to images with a mermaid filter. Then verify: total word count and page count are in range, all diagrams render, the chapter structure is intact, and — critically — the private profile file is NOT present in the compiled artifact.


Definition of done

  • The user gave explicit, source-by-source consent and chose the tone.
  • A private, git-ignored profile.md exists and drives the chapters.
  • ~30-40 chapters exist, each following the full template, totaling ~120k+ words.
  • PDF and EPUB build cleanly with diagrams rendered.
  • The compiled book contains zero raw private content and excludes the profile.

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.