Skip to content

Humane documentation that machines can parse

License

Notifications You must be signed in to change notification settings

pbierkortte/Engram

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2 Commits
 
 
 
 
 
 

Repository files navigation

Engram: Humane documentation that machines can parse

Summary

Engram extends familiar Markdown patterns with lightweight conventions for structured data. If you can write Markdown, you can write Engram.

What is Engram?

Engram is for technical writers, worldbuilders, prompt engineers, and developers who refuse to choose between readable prose and structured data.

It exists because current formats fail them: JSON and YAML serve machines first, Markdown serves humans but can't be reliably parsed, and maintaining both means duplicating everything.

It works by treating context as the parser's guide. Wrap documents in --- name and --- boundaries, then write normal Markdown. A blank line before Key: Value triggers data mode; without it, that same text stays prose. Lists attach to their preceding key or label. Four-space indentation continues values across lines. If a human reads it naturally, the machine parses it correctly.

It produces documents that humans scan as documentation while machines extract as structured data—version-control friendly, portable across any Markdown renderer, readable and programmable from the same source file.


Quick Example

--- character-profile

# Character Profile: John Doe

A senior engineer who combines technical skill with leadership

## Basic Information

Name: John Doe
Age: 32
Occupation: Lead Software Engineer

Skills:
- Programming
- Technical Writing
- System Analysis

---

This natural documentation is also fully machine-parseable with clear structure for automated processing.


Use Cases

Engram is a Markdown-native YAML alternative that excels at:

RPG Worldbuilding/Lore Bibles:

  • Write stories and character bios where game engines can parse stats directly from the narrative text
  • Seamless integration of lore and mechanics

Software Architecture Decision Records (ADRs):

  • Mix high-level reasoning (prose) with specific metadata (status, deciders, date)
  • Version control friendly

LLM Prompt Engineering:

  • Store prompts where system instructions are prose but parameters are structured data
  • Easy to read, easy to parse

Other Applications:

  • Technical documentation with embedded metadata
  • Configuration files that humans actually need to read
  • Knowledge bases mixing narrative and structured information
  • API documentation with parseable examples
  • Product specifications where context and data coexist

Documentation

See ENGRAM_SPEC.engram for the complete specification.


License

GPL v3 - See LICENSE for details.

About

Humane documentation that machines can parse

Topics

Resources

License

Stars

Watchers

Forks