prompt

Free

Design constraint-based workflows for reliable LLM planning

FreeFree tier

Type

Open Source

Visit Website GitHub (40 stars)

About prompt

The Constraint Typology Architect is a specialized system prompt from the awesome-prompts collection, designed to guide LLMs in constructing constraint-based workflows for planning and decision-making systems. Based on the U-Define framework, it instructs the LLM to elicit constraints from users in natural language and classify them into two categories: hard rules that must never be violated and soft preferences that allow flexibility. The prompt emphasizes falsifiability, external auditable verification for hard constraints, and contextual evaluation for soft constraints, avoiding the pitfalls of brittle hard-only rules or confusing numeric weights. It supports incremental refinement as domain understanding evolves, ensuring user intent is captured reliably without over-constraining or losing nuance.

Key Features

Elicits constraints in natural language without requiring numeric weights or priorities

Classifies constraints as hard (must never violate) or soft (flexible preferences) with matched verification methods

Ensures each constraint is falsifiable by asking clarifying questions about violation scenarios

Requires external, auditable verification for hard constraints and judgment-based evaluation for soft constraints

Supports incremental refinement of constraints without invalidating prior plans

Prevents mixing verification styles and collapsing constraints into single numeric weights

Pros & Cons

Pros

Clearly separates hard and soft constraints to avoid brittle over-constraining
Eliminates confusing numeric flexibility weights that produce unpredictable trade-offs
Promotes falsifiable constraint elicitation for actionable specifications
Uses external verification for hard constraints, increasing trust and auditability
Allows incremental refinement without disrupting previously validated plans
Grounded in academic research (U-Define paper) for rigor

Cons

Requires manual user interaction for initial constraint elicitation; not fully autonomous
Assumes constraints expressible in natural language, limiting formal specification
May need multiple refinement rounds to capture all domain nuances
No built-in execution or planning engine; only provides the prompt structure

Best For

Designing workflow for LLM-based planning systems in autonomous agentsConfiguring code generation or decision systems with user-defined rulesRequirements analysis for systems where constraints evolve over timeBuilding auditable verification pipelines for safety-critical AI applicationsTeaching LLMs to handle both rigid rules and flexible preferences in interactive settings

FAQ

What is the Constraint Typology Architect prompt?

It is a system prompt from the awesome-prompts GitHub repository that instructs an LLM to act as a constraint typology architect. Its job is to design constraint-based workflows for LLM planning systems by eliciting user constraints in natural language and classifying them as hard (must never violate) or soft (flexible preferences).

How does the prompt classify constraints?

Hard constraints are rules that make the plan unacceptable if violated, regardless of context; they require exhaustive, sound verification. Soft constraints are preferences that allow flexibility and require contextual, judgment-based evaluation. The prompt avoids numeric weights or priority scores.

What is the source of this prompt?

The prompt is based on the U-Define framework from the paper 'U-Define: Designing User Workflows for Hard and Soft Constraints in LLM-Based Planning' (arXiv 2605.02765, May 2026). It is hosted in the ai-boost/awesome-prompts GitHub repository.

Can this prompt be used for any LLM?

Yes, the prompt is designed to be used with any LLM that supports system-level instructions. It treats the LLM as a black box planner, and the verification methods are external and auditable.