theta-spec - a humble harness agnostic configuration spec
Another config standard for AI agents when each harness already has its own format.
A specification pattern for consistent AI code generation.
Turns fuzzy AI prompts into a lightweight contract: SIGNATURE, BEHAVIOR (WHEN/THEN) and TESTS can be written as .rune YAML or embedded Markdown, letting any model or language generate the same behavior. Clever, low-friction idea with practical utilities (drift detection, reverse-engineer skills), but it’s a pattern rather than a product — adoption will hinge on tooling and CI integrations to make these specs enforceable at scale.
Developers, engineering teams, and prompt/AI-tooling engineers who use AI-assisted code generation
It's a pattern, not a framework. Works as YAML files or Markdown sections. Any language, any AI tool, no installation.
Includes 7 skills (markdown files you load into your AI tool) for the lifecycle: create specs, validate, refine, generate tests, detect drift, reverse-engineer from code, multi-language generation.
This is the GitHub project: https://github.com/vict00r99/Rune-stone
Another config standard for AI agents when each harness already has its own format.
ANML proposes a new protocol for agents, but adoption depends on an unproven agentic web.
Up-to-date LSP 3.17 support in Go beats outdated alternatives like sourcegraph/go-lsp.
Spec-as-code is clever, but multi-target compilation claims need proof at v0.1.
Five SQLite engines from one spec, all passing 99% of the upstream sqllogictest corpus.
Locks brand styles for AI images, but Canva and Adobe already do this.