The State Engine Framework is a collection of high-level templates that turn messy AI chats into structured, repeatable workflows.
Instead of:
- scrolling through 200+ messages,
- re-explaining context every time,
- watching models randomly rewrite your whole document,
SEF gives you State Templates: copy-paste blocks you drop at the start of a conversation to turn the AI into a constrained engine:
- PST – Project State Template → For multi-section projects (articles, specs, features, chapters, etc.)
- SST – Search State Template → For precise, Google-aware search sessions (what you search, how, and in what format).
More templates (coding, writing, data, research, trading, etc.) will live in this repo as the framework evolves.
LLMs are powerful, but:
- They don’t remember your structure.
- They over-rewrite when you only wanted a small tweak.
- They force you to scroll through a flood of text to see what changed.
- They mix “conversation” with “project work” in a way that’s painful for devs and writers.
State Templates fix that by:
- Freezing a project/search into sections.
- Defining dependencies (what affects what).
- Controlling what can change and how updates propagate.
- Generating Output Templates you can reuse in future sessions.
You’re not just prompting anymore.
You’re designing State Engines.
core/
PST_v1.1.md # Project State Template
SST_v1.0.md # Search State Template
examples/
pst_article_example.md # Example of PST used for a Dev article
sst_search_example.md # Example of SST used for a focused search
meta/
DESIGN_NOTES.md # Philosophy & internals
CONTRIBUTING.md # How to propose new State Templates
ROADMAP.md # Planned templates & evolutions