qikly
Active·★ 17·Apache-2.0·Updated 2026-10-19
qikly addresses the problem of AI writing both code and its tests, which compromises test validity. It uses two distinct agents: one generates tests from acceptance criteria, while the other writes code without ever seeing those criteria, ensuring independent verification.
#AI Testing#Code Generation#LLM Agents#Python#Test Automation#Acceptance Testing#Independent Verification
01
Features
01Withheld Acceptance Criteria: The coding agent never sees the acceptance criteria, ensuring test independence and preventing tests from being shaped by the implementation.
02Staged Test Generation: Integration and system tests are written before any code exists, followed by unit tests, ensuring the standard cannot be shaped by the thing it judges.
03Durable Executable Test Suites: Produces standard pytest files and JUnit XML, which are persistent assets that can be kept, read, and integrated into CI/CD pipelines.
04Comprehensive Audit Trail: Records every test run, fix, patch, and application outcome in an append-only log, along with HTML timeline and metrics reports for full reproducibility.
05Criteria Management Tools: Includes tools to draft, validate, compare, and import acceptance criteria from various sources (e.g., tickets, Gherkin files), aiding in defining the 'standard'.
02
Compatibility
Python 3.10
Runtime
Verified via docs
Python 3.11
Runtime
Verified via docs
Python 3.12
Runtime
Verified via docs
Python 3.13
Runtime
Verified via docs
MacOS
OS
Verified via docs
GNU patch
Dependency
Verified via docs
03
Quick start
1
$ pip install qikly
04
Use cases
↳AI-driven Development for Data Transformation: Generating code and independent tests for self-contained Python modules performing ETL, data merges, calculations, or validation routines.
↳Independent Verification in Critical Domains: Ensuring objective and independent verification for applications in highly regulated industries such as automotive, medical devices, fintech, and defense.
↳Verifying Third-Party or Legacy Code: Supplying an existing implementation and generating an independent test suite from acceptance criteria, without the test agent ever seeing the provided code.
↳Starting Development from Specification: Generating an initial implementation and a justifying test suite directly from a specification, without any pre-existing code.
05
Alternatives
DeepCode★ 16.7k
DeepCode is an open-source coding agent from HKUDS for real software engineering: goal-driven loops, evidence-based verification, durable sessions, and Paper2Code research reproduction.
mini-swe-agent★ 8.7k
The 100 line AI agent that solves GitHub issues or helps you in your command line. Radically simple, no huge configs, no giant monorepo—but scores >74% on SWE-bench verified!
ai-rulez★ 163
Standards-compliant lifecycle tool for agent knowledge and capabilities: author, generate, bundle, validate, govern, publish.
unity-api-mcp★ 67
Instant, accurate Unity API lookups instead of expensive source file reads, saving your agent tokens, context, and hallucinations
ios-agent-skill★ 39
Swift source, Apple guides and MCP tools for AI coding agents. Review iOS code, scaffold apps, and build/run/preview with Xcode Simulator.
Comments
Log in to leave a comment
No comments yet. Be the first!