# Intent engineering

The practice of turning goals into versioned specifications, constraints, interfaces, invariants, decision rights, acceptance criteria, and testable outcomes before agents implement them.

core-concept · Foundations · maturing · Reviewed 2026-08-09

## Definition

The practice of turning goals into versioned specifications, constraints, interfaces, invariants, decision rights, acceptance criteria, and testable outcomes before agents implement them. A prompt requests work. A durable intent system defines what correct, safe, and complete work means.

Spec-driven practice is growing, but spec-as-source is not a universal consensus.

## Why it matters

A prompt requests work. A durable intent system defines what correct, safe, and complete work means.

The operating shift is from “Prompt as instruction” to “Specification as executable intent.”

## System anatomy

- **Practice 1:** Version requirements beside implementation
- **Practice 2:** Encode invariants and acceptance criteria
- **Practice 3:** Trace design decisions back to intent



## Important distinctions

- **Prompt as instruction:** This concept moves the engineering system toward specification as executable intent.



## Implementation signals

- Version requirements beside implementation
- Encode invariants and acceptance criteria
- Trace design decisions back to intent

## Failure modes

- Verbose specifications without enforceable tests
- Assuming generated implementation keeps the specification current



## Related knowledge

- [Context engineering](https://aisdlc.ai/agentic-engineering/context-engineering) — The deliberate selection and maintenance of instructions, knowledge, tools, state, examples, and artifacts within a finite model attention budget.
- [Eval-driven development](https://aisdlc.ai/agentic-engineering/eval-driven-development) — An engineering loop that converts expected behavior and observed failures into repeatable evaluations combining deterministic checks, environment inspection, security testing, model graders, repeated trials, and human judgment.
- [Evidence engineering](https://aisdlc.ai/agentic-engineering/evidence-engineering) — The design of versioned, queryable evidence linking requirements, decisions, implementations, tests, evaluations, approvals, deployments, runtime signals, and lifecycle actions.

## Sources and further study

- [GitHub — Spec-driven development with AI](https://github.blog/ai-and-ml/generative-ai/spec-driven-development-with-ai-get-started-with-a-new-open-source-toolkit/) — A structured specification-to-plan-to-task workflow for making intent durable and actionable by coding agents. **Use:** First-party case study. **Limitation:** This first-party account documents one organization, product, or implementation context and should not be generalized without local evidence.
- [GitHub Spec Kit — Spec persistence models](https://github.github.com/spec-kit/concepts/spec-persistence.html) — Distinguishes spec-first, spec-anchored, and spec-as-source approaches without asserting one universal model. **Use:** First-party case study. **Limitation:** This first-party account documents one organization, product, or implementation context and should not be generalized without local evidence.
- [OpenAI — Harness engineering: leveraging Codex in an agent-first world](https://openai.com/index/harness-engineering/) — A production case study in designing repositories, tools, feedback loops, and environments for agent-first delivery. **Use:** First-party case study. **Limitation:** This first-party account documents one organization, product, or implementation context and should not be generalized without local evidence.

---

This library synthesizes cited research, standards, official documentation, and clearly attributed practitioner perspectives. Maturity describes the state of a concept—not vendor endorsement, production readiness, or permission to deploy. Benchmarks and demonstrations do not replace use-case evaluation, governed controls, independent verification, or named human release authority.