Skip to main content
IntentLang
Draft documentation. Syntax and behavior are illustrative and will change before v1.

The IntentLang Manifesto

IntentLang is the intent language for AI-era software. It lets engineers define what software should do, why it matters, what must never happen, and how the result must be verified before code is generated, changed, or shipped.

Prompt is temporary. Intent is durable.

A prompt is a conversation. Consider:

Build a secure invoice creation flow that prevents duplicates, audits every invoice, and never logs payment tokens.

That is useful, but it is not durable. It is not a contract. It is not cleanly version-controlled. It is not automatically verifiable.

The prompt should become a .intent file: reviewable, versionable, testable, explainable, compilable, verifiable, and portable across target languages. That is far more valuable than the prompt, because the intent is what the team actually commits to.

The path from purpose to proof

Prompt → Intent → Contract → Plan → Implementation → Verification → Proof
  • Prompt is how the conversation starts.
  • Intent is what the team commits to.
  • Code is how the system fulfills it.
  • Proof is how trust is earned.

Expanded, the pipeline is:

Human prompt / requirement / architecture discussion
  → IntentLang .intent file
  → Contract graph
  → Architecture graph
  → Implementation plan
  → Target artifacts
  → Verification
  → Proof

Intent becomes many kinds of artifact

Implementation is the output, not the point. Target artifacts can include:

  • TypeScript, Python, C# / .NET, Java, Go, Rust code
  • OpenAPI specs
  • Mermaid diagrams
  • Markdown docs
  • Test plans and unit test scaffolds
  • Security rules
  • OpenThunder contracts
  • Repo Mastery learning paths
  • SkillsTech Talk defense prompts
  • Proof JSON

So object-oriented code is one possible target, not the whole story.

IntentLang sits above programming paradigms. It can target object-oriented, functional, service-oriented, event-driven, API-first, or infrastructure code depending on the adapter.

Intent becomes target-specific implementation using the idioms of each ecosystem.

What IntentLang is not

  • Not just a prompt format.
  • Not magic AI code generation.
  • Not a replacement for TypeScript, Python, Java, .NET, Rust, or Go.
  • Not production-ready. This is early, honest work.

The right framing:

IntentLang turns software purpose into explicit contracts, architecture, verification, and proof, then uses compilers and optional AI to help produce target implementations people can trust.

AI-era, not AI-magic

The deterministic compiler must work without AI. Every optional AI action produces metadata: provider, model, prompt hash, input and output hashes, verification result, and human approval status. Humans approve, verify, and own the result.

Intent Drift

Intent drift happens when the implementation no longer satisfies the declared mission, guarantees, architecture, constraints, or verification rules. For example: a guarantee exists but no test verifies it, a never rule is violated, a service publishes an event not declared in intent, a public API changes but the intent was not updated, or a field marked Secret is logged.

Detecting and closing drift, later via OpenThunder, is a core reason IntentLang exists. Generating code is cheap; keeping code faithful to intent is the hard, valuable part.

The proof chain

Proof is the through-line of the SkillsTech ecosystem. Each product proves a different thing about the same mission:

  • SkillsTech Compiler proves the language can produce artifacts.
  • OpenThunder proves the repo matches the declared intent.
  • Repo Mastery proves the human understands the mission.
  • SkillsTech Certified proves the learner understands the method.

Together they turn a mission into compounding evidence: the artifact exists, the implementation still matches it, the engineer who owns it understands it, and the practitioner has proven the method itself. That is how trust is earned in AI-era software, not by trusting the code, but by proving the intent behind it.

Syntax overview →