Code Completion Semantic Constraints Agent Skill

Code Completion with Semantic Constraints

Overview

Complete partial code snippets while satisfying specified semantic constraints. Produces compilable code, verification tests, and a detailed report explaining how each constraint was satisfied.

Workflow

1. Parse Input

Extract and categorize the provided information:

Partial Code: Identify the incomplete code structure (function signature, class skeleton, method stub, etc.)

Semantic Constraints: Categorize constraints by type:

• Type constraints: Variable types, return types, generic bounds
• Invariants: Pre-conditions, post-conditions, loop invariants
• Behavioral constraints: Expected input/output pairs, edge case handling
• Interface contracts: Method signatures, protocol conformance
• Performance constraints: Time/space complexity requirements

2. Analyze Constraints

For each constraint:

• Determine if it's satisfiable
• Identify dependencies between constraints
• Note any conflicts or ambiguities
• If constraints are unclear or conflicting, ask for clarification before proceeding

3. Complete the Code

Generate code that:

• Compiles without errors in the target language
• Satisfies all specified constraints
• Follows language idioms and best practices
• Includes necessary imports, type annotations, and error handling
• Uses minimal complexity (avoid over-engineering)

4. Generate Verification Tests

Create minimal test cases that verify:

• Type constraints are respected
• Pre-conditions and post-conditions hold
• Expected input/output behavior is correct
• Edge cases are handled properly
• Performance constraints are met (if specified)

5. Produce Constraint Satisfaction Report

Document how each constraint was satisfied:

• Map each constraint to the code that satisfies it
• Explain the reasoning for implementation choices
• Note any assumptions made
• Highlight any constraints that required trade-offs

Examples

Example 1: Type and Behavioral Constraints

Input:

def process_items(items):
    # TODO: complete this function
    pass

Constraints:

• items is a list of integers
• Return a list of unique integers in ascending order
• Handle empty list (return empty list)
• Time complexity: O(n log n)

Output: Completed function with sorting logic, tests for empty/normal/duplicate cases, and report explaining constraint satisfaction.

Example 2: Interface Contract

Input:

public class DataCache implements Cache {
    // TODO: implement required methods
}

Constraints:

• Implement Cache interface (get, put, remove methods)
• Thread-safe operations
• LRU eviction policy with max size 100
• Return null for missing keys

Output: Complete class implementation, concurrency tests, and report mapping each interface method to implementation.

Example 3: Invariant Preservation

Input:

void update_balance(Account* acc, int amount) {
    // TODO: complete
}

Constraints:

• Pre-condition: acc != NULL && acc->balance >= 0
• Post-condition: acc->balance >= 0 (balance never negative)
• If amount would make balance negative, set to 0 instead

Output: Implementation with bounds checking, tests for edge cases, and report showing invariant preservation.

Output Format

Provide three components:

1. Completed Code

• Fully compilable and executable
• Includes necessary imports, type annotations, error handling
• Follows language conventions

2. Verification Tests

• Minimal test suite covering constraint verification
• Include edge cases and boundary conditions
• Use appropriate testing framework for the language

3. Constraint Satisfaction Report

• Table or list mapping each constraint to implementation details
• Explanation of design choices
• Any assumptions or trade-offs made

Language Support

This skill works with any programming language. Adapt constraint types and verification approaches to language-specific features (e.g., type systems, contract programming, assertion libraries).