agentleFS
MaGe1993MaGe1993/Claude-Code-Base-Harness/.claude/skills/masterprompt/SKILL.md

masterprompt

Skill1 starsChanged 5 months agoMITView on GitHub

Generate a maximal, comprehensive masterprompt on any topic — research the domain, then build the most complete expert prompt: persona, sub-domains, edge cases, behavioral mandates.

The file

We split this skill into the rules it gives an agent

9 rules. About 0 are common, 8 are distinctive, and 1 is 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.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Always research the topic before writing — never build from memory alone.

    line 27
  2. Distinctive:

    Never truncate or abbreviate the output.

    line 28
  3. Distinctive:

    Never add meta-commentary, preamble, or explanation outside the prompt itself.

    line 29
  4. Distinctive:

    Never ask the user for clarification — interpret the topic maximally and begin.

    line 30
  5. Distinctive:

    Before writing a single line of the prompt, research the topic across all relevant dimensions.

    line 34
  6. Very specific:

    Use WebSearch and WebFetch to gather current, specific information.

    line 46
  7. Distinctive:

    All key sub-topics, each with a sentence of substance

    line 80
  8. Distinctive:

    Use of examples, analogies, counter-examples

    line 100
  9. Distinctive:

    All six sections from §5, in order

    line 116

≈ “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

Masterprompt

1§ Purpose

Generate a comprehensive, maximal masterprompt for any given topic. The output is a standalone, ultra-detailed prompt that transforms Claude into a deep expert on that exact subject. No topic is too specific or too abstract — truck driving, three yellow chickens, quantum mechanics, medieval baking: every topic gets the same maximum treatment.

A masterprompt is not a summary. It is the densest possible expert briefing: a prompt so complete that anyone reading it walks away knowing exactly what questions to ask, what answers to expect, and what a true master of the topic would know.

2§ Use when

  • User asks for a masterprompt, mega-prompt, super-prompt, or expert prompt on any topic.
  • User wants a comprehensive prompt covering all aspects of a subject.
  • User names a topic and wants the most powerful possible prompt for it.

3§ Hard constraints

  • Always research the topic before writing — never build from memory alone.
  • Never truncate or abbreviate the output. Length is a feature, not a flaw.
  • Never add meta-commentary, preamble, or explanation outside the prompt itself.
  • Never ask the user for clarification — interpret the topic maximally and begin.
  • The output IS the prompt. Deliver it ready to paste.

4§ Research phase

Before writing a single line of the prompt, research the topic across all relevant dimensions. For each topic, identify:

  • Core domain and sub-domains: what are the main areas of knowledge?
  • Expert vocabulary: what terms, jargon, classifications does a master use?
  • Practical application: what does this look like in real-world use?
  • Edge cases and rare scenarios: what does a novice miss but an expert knows?
  • Regulatory, safety, or ethical dimensions: what constraints apply?
  • Cultural, historical, or contextual background: what shaped this field?
  • Common misconceptions: what do beginners get wrong?
  • Hierarchies and systems: how is this domain structured internally?

Use WebSearch and WebFetch to gather current, specific information. Research depth should match topic complexity — more layers for professional domains, less for narrow/playful topics (but even "3 yellow chickens" gets full creative treatment with breed specifics, behavior, care, lore, etc.).

5§ Prompt construction

Build the masterprompt in this structure. All sections are mandatory unless the topic makes a section genuinely inapplicable.

5.1§ Persona block

Open with a richly detailed expert identity. Include:

  • Role title and seniority level
  • Years and type of experience
  • Specific sub-specializations within the domain
  • Real-world context (where they work, what they've done)
  • Mindset and approach to the subject

5.2§ Knowledge domains

Enumerate all major knowledge areas the expert holds. Cover:

  • Primary domain knowledge in full depth
  • Adjacent domains that intersect
  • Technical, theoretical, and practical knowledge
  • Tools, instruments, terminology, classifications used

5.3§ Behavioral mandates

Define how the expert thinks and communicates:

  • How they analyze problems in this domain
  • What precision standard they apply
  • How they handle uncertainty or conflicting information
  • What they refuse to simplify beyond a certain point
  • How they distinguish beginner from expert questions

5.4§ Domain specifics

The densest section — everything specific to this topic:

  • All key sub-topics, each with a sentence of substance
  • Standard processes, sequences, or protocols
  • Known failure modes, risks, and how to handle them
  • Best practices vs. common practice vs. expert shortcuts
  • Relevant standards, regulations, certifications, or norms (where applicable)
  • Historical development and current state of the field
  • Open questions, debates, or unsettled areas in the domain

5.5§ Interaction style

Define how the masterprompt persona responds:

  • Tone (e.g., direct, technical, patient, authoritative)
  • Default detail level
  • How they handle vague questions (interpret maximally, ask once if critical)
  • Examples: when they use them, at what depth
  • What they proactively add beyond what was asked

5.6§ Output format defaults

Define what the expert's default output looks like:

  • Structure (prose, lists, hierarchies, tables — when to use which)
  • Length defaults per question type
  • Use of examples, analogies, counter-examples
  • Caveats: when included, when omitted

6§ Length and density

The masterprompt must be substantial. A minimal topic warrants at minimum one full page of dense content. A complex professional domain warrants 3–6 pages. Density over filler: every sentence carries specific, non-obvious information.

Filler phrases that consume tokens without adding knowledge ("it is important to", "one must consider", "always keep in mind") are forbidden. Replace with the actual content those phrases were gesturing at.

7§ Output contract

The output is the masterprompt itself — nothing else.

Must include:

  • All six sections from §5, in order
  • Specific, researchable facts woven throughout (not generic expert-speak)
  • Expert vocabulary used correctly and in context
  • Minimum one full page of content for any topic

Must not include:

  • Meta-text explaining what the prompt does
  • Preamble ("Here is your masterprompt:")
  • Trailing commentary or instructions to the user
  • Apologies, caveats about AI limitations, or requests for feedback
  • Filler sentences that state no specific fact

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.