Skip to content
TCAFTask-Contract AI Development Framework
TCAF — Task-Contract AI Development Framework

TCAF 0.4.0 · Public preview

Use AI to build software. Keep engineering control.

TCAF is a governance-first, spec-driven framework for controlled AI software delivery. It implements Task-Contract Development: an approach that integrates spec-driven planning, agentic/harness execution and software delivery governance into explicit, bounded and verifiable units of change.
Clear constraints for AI. Control and flexibility for the developer.

Bound the task

Turn a request into an explicit Task Contract before consequential implementation begins.

Preserve the codebase

The AI works with existing architecture, behavior and local patterns instead of expanding scope or rewriting opportunistically.

Verify the result

Executed checks, pending verification and limitations stay visible before the developer accepts the work.

Where TCAF fits

Three disciplines converge in Task-Contract Development.

TCAF brings them together in one practical delivery workflow while developers retain authority over architecture, code and acceptance.

Spec-Driven Development

Specifications and repository evidence define what should be built.

Agentic / Harness Engineering

Runtime, adapters, context and rules make AI execution repeatable across tools.

Software Delivery Governance

Planning, Task Contracts, sync, verification and approval keep consequential decisions developer-owned.

Inside one task

A short feedback loop, not autonomous project ownership.

01

Request

Developer defines the intended outcome.

02

Contract

TCAF makes scope, preservation and acceptance explicit.

03

Execute

The AI can execute the approved work inside that boundary.

04

Verify

Evidence is reported before developer acceptance.

The v0.4.0 workflow

Five operations. Use only the one you need next.

Need the reasoning behind the workflow?

See how TCAF works →