agentleFS
BotTony329BotTony329/consulting-deck-skill/CLAUDE.md

consulting-deck-skill

CLAUDE.md1 starsChanged 3 months agoMITView on GitHub

Build .pptx decks that argue instead of describe. This file is the short form; the full method lives in SKILL.md and references/. Read those before building anything non-trivial. Do not start placing slides until you know: audience, the decision the deck supports, the core question it answers, the final message, the evidence standard that audience demands, and the visual identity. Ask about the theme rather than picking one silently — a board paper for an insurer and a research report…

  • Installs packages

The file

We split this CLAUDE.md into the rules it gives an agent

4 rules. About 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    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:

    Do not start placing slides until you know: audience, the decision the deck supports, the core question it answers, the final message, the …

    line 19
  2. Distinctive:

    Ask about the theme rather than picking one silently — a board paper for an insurer and a research report for a school should not look …

    line 23
  3. Very specific:

    Every figure carries a source line: Never invent one. If a number has no source, cut it or mark it DATA REQUIRED.

    line 39
  4. Distinctive:

    Put another company's logo, wordmark, tagline, licensed photography, or copied layout compositions into a deck for someone who doesn't own …

    line 96

≈ “Common” is an estimate: at least 1 in 5 of the 27,641 CLAUDE.md 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
CLAUDE.md
<!-- Generated from AGENTS.md by scripts/sync_agent_files.py. Edit AGENTS.md, not this file. -->

consulting-deck — agent operating guide

Build .pptx decks that argue instead of describe. This file is the short form; the full method lives in SKILL.md and references/. Read those before building anything non-trivial.

Install

pip install -e .            # or: pip install consulting-deck
consulting-deck themes      # list the visual identities
consulting-deck schema      # list the slide types and their fields

Before you build, ask

Do not start placing slides until you know: audience, the decision the deck supports, the core question it answers, the final message, the evidence standard that audience demands, and the visual identity.

Ask about the theme rather than picking one silently — a board paper for an insurer and a research report for a school should not look alike. Offer the shipped themes and ask whether the user has a brand accent colour. Unattended, choose one, say which and why, and note that a brand colour is a one-line swap.

The method, in eight rules

  1. Research first. Gather facts, figures and citations before opening the tooling. A beautiful deck on invented numbers is worthless.
  2. Write the storyline as text first — one line per slide, each the message of that slide, not its topic. If the lines don't argue toward the recommendation on their own, the deck won't either.
  3. Titles assert, they don't label. "Pandemic accelerated consolidation" beats "Pandemic impact". A reader flipping only the assertions should get the argument.
  4. Push every number down the so-what ladder until it changes a decision. Description is the first rung; the third is where the fee is earned.
  5. Every figure carries a source line. Never invent one. If a number has no source, cut it or mark it DATA REQUIRED.
  6. Keep claim types distinct: fact / inference / hypothesis / recommendation. Restructure supplied material aggressively; never silently invent.
  7. One number per section gets a whole slide at 190pt. It is what people remember.
  8. One message per slide, ≤5 supporting lines. A paragraph over ~80 characters wants to be a chart, a table, or an appendix page.

Build

Spec route — for any agent that can write a file and run a command:

# deck.yaml
lang: en
theme: slate
slides:
  - {type: title, title: Market entry study, subtitle: RTD tea · China}
  - {type: section, number: 1, title: Market size, scope: What the category looks like}
  - type: content
    category: Industry status
    message: The category is large, growing fast, and priced down.
    body: ["Revenue reached ¥181bn in 2022, up 108% since 2018"]
    source: CIC, Xinhua
    body_width: 5.4
    chart: {kind: column, categories: ["2018","2022"], values: [870,1810],
            at: [6.6,2.75,5.2,3.2], unit: ¥100m}
  - {type: stat, number: "108", unit: "%", caption: growth over four years, source: CIC}
  - {type: closing, org: Your organisation}
consulting-deck build deck.yaml -o deck.pptx --check

Python route — full control:

from consulting_deck import Deck
d = Deck(lang="en", theme={"base": "slate", "accent": "#7A1FA2"})
d.title_slide("Market entry study", "RTD tea · China")
d.big_stat("108", "growth over four years", unit_suffix="%", source="CIC")
d.save("deck.pptx")

Verify — all three gates

consulting-deck check deck.pptx     # off-canvas text, overflow, missing sources
consulting-deck titles deck.pptx    # read the assertions as prose: do they argue?
soffice --headless --convert-to pdf deck.pptx && pdftoppm -png -r 70 deck.pdf p

The rendered output is the truth, not the code that produced it. Look at the images.

Never

Put another company's logo, wordmark, tagline, licensed photography, or copied layout compositions into a deck for someone who doesn't own them. Method transfers between decks; visual identity does not.

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.