agentleFS
samulee003samulee003/agentic-ppt-skills/AGENTS.md

agentic-ppt-skills

AGENTS.md0 starsChanged 24 days agoMITView on GitHub

This file is the machine-oriented entry point for agents operating from this repository. The root README.md is the human-oriented guide. 1. If the current directory is the user's project repository, install there. 2. If it is this skills repository, a scratch directory, or no repository, create a separate deck workspace and tell the user its path. 3. Choose exactly one runtime directory: .agents/skills/ for most runtimes, or .claude/skills/ for Claude Code. Copy skills and adapters to the matching directory; do…

The file

Show its rules and text

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

13 rules. About 0 are common, 8 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈8

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If the current directory is the user's project repository, install there.

    line 8
  2. Distinctive:

    If it is this skills repository, a scratch directory, or no repository, create a separate deck workspace and tell the user its path.

    line 9
  3. Very specific:

    If both directories contain different make-presentation/SKILL.md files, stop and ask the user which one to target.

    line 14
  4. Very specific:

    Before continuing, verify Node 18+, the orchestrator, the status helper, and at least one adapter (open-slide, genoffice, marp, or …

    line 28
  5. Distinctive:

    Stop and report any failed check.

    line 39
  6. Distinctive:

    Before starting, ask one question and wait:

    line 44
  7. Distinctive:

    Do not ask for mode or engine yet.

    line 49
  8. Very specific:

    Run exactly one focused skill for the earliest unresolved gate.

    line 63
  9. Distinctive:

    Create each named artifact yourself from its template.

    line 65
  10. Very specific:

    Never hand-edit STATUS.md.

    line 68
  11. Very specific:

    Do not claim success until next reports complete.

    line 80
  12. Distinctive:

    every exported file path

    line 83
  13. Distinctive:

    Do not skip evidence, readability, or export checks.

    line 86

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.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
AGENTS.md

Instructions for AI agents

This file is the machine-oriented entry point for agents operating from this repository. The root README.md is the human-oriented guide.

1. Detect the install target

  1. If the current directory is the user's project repository, install there.
  2. If it is this skills repository, a scratch directory, or no repository, create a separate deck workspace and tell the user its path.
  3. Choose exactly one runtime directory: .agents/skills/ for most runtimes, or .claude/skills/ for Claude Code. Copy skills and adapters to the matching directory; do not symlink.
  4. If both directories contain different make-presentation/SKILL.md files, stop and ask the user which one to target.

2. Install and verify

Set SRC to the actual path of this repository; do not assume its folder name.

SRC=/absolute/path/to/agentic-ppt-skills
mkdir -p .agents/skills .agents/adapters
cp -r "$SRC/skills/." .agents/skills/
cp -r "$SRC/adapters/." .agents/adapters/

Before continuing, verify Node 18+, the orchestrator, the status helper, and at least one adapter (open-slide, genoffice, marp, or pptxgenjs):

node --version                                                # must be v18 or higher
test -f .agents/skills/make-presentation/SKILL.md && echo "orchestrator present"
test -f .agents/skills/make-presentation/scripts/status.mjs && echo "status helper present"
node .agents/skills/make-presentation/scripts/status.mjs 2>&1 | head -1  # "error: Expected command..." confirms it ran
ls .agents/adapters/open-slide/ .agents/adapters/genoffice/   # must list adapter files

Stop and report any failed check. Do not install dependencies for the zero-dependency status helper.

3. Ask for the source

Before starting, ask one question and wait:

What is the source for this presentation? It can be notes, an existing deck, a URL, or a topic. Default: notes in the current folder.

Do not ask for mode or engine yet. Mode is evaluated after source intake and the engine is selected at the prototype gate.

4. Run the orchestrator

Initialize the tracker, then load make-presentation/SKILL.md:

node .agents/skills/make-presentation/scripts/status.mjs \
  init presentation-work/<deck-id> <deck-id> <source-root>
node .agents/skills/make-presentation/scripts/status.mjs \
  next presentation-work/<deck-id>

Run exactly one focused skill for the earliest unresolved gate. SKILL.md is the agent contract; skills/SKILLS-INDEX.md is the human-facing catalog. Create each named artifact yourself from its template. Never skip a gate because an output already exists.

Never hand-edit STATUS.md. Use only init, next, result, invalidate, set-mode, set-engine, and summary from the status helper. Mark a gate passed only with real --artifact paths.

At the prototype gate, ask the user to choose an installed adapter (default: open-slide, or genoffice when generating a native byte-preserving Office .pptx via GenOffice), verify its directory, record it with set-engine, and read that adapter's README.md and prototype.md. Production QA inherits the choice and must also read qa.md.

5. Completion contract

Do not claim success until next reports complete. Then provide:

  • presentation-work/<deck-id>/STATUS.md
  • every exported file path
  • the pipeline work directory

Do not skip evidence, readability, or export checks. If the environment or a verification step is unclear, stop and ask instead of guessing.

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.