Metadata-Version: 2.4
Name: assaylab
Version: 0.0.1
Summary: Validation intelligence for CI — attested test-selection with verifiable confidence bounds, failure-signature RCA, and analytics. Contract-compatible with agentsensory.
Project-URL: Homepage, https://github.com/amitpatole/assaylab
Project-URL: Repository, https://github.com/amitpatole/assaylab
Author-email: Amit Patole <amit.patole@gmail.com>
License: MIT
License-File: LICENSE
Keywords: agentsensory,attestation,ci,flaky-tests,root-cause-analysis,test-selection,testing,validation
Classifier: Development Status :: 1 - Planning
Classifier: Intended Audience :: Developers
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3.11
Classifier: Programming Language :: Python :: 3.12
Classifier: Topic :: Software Development :: Quality Assurance
Classifier: Topic :: Software Development :: Testing
Requires-Python: >=3.11
Provides-Extra: dev
Requires-Dist: mypy>=1.10; extra == 'dev'
Requires-Dist: pytest>=8.0; extra == 'dev'
Requires-Dist: ruff>=0.5; extra == 'dev'
Description-Content-Type: text/markdown

# assaylab

**Validation intelligence for CI.** Attested test-selection with verifiable
confidence bounds, failure-signature root-cause analysis, and analytics that
support data-driven engineering decisions.

> **Status:** name reserved, implementation in progress.

## What it does (planned)

**Validation intelligence & analytics**
- Failure-signature clustering over historical CI/test results.
- Automated root-cause analysis and flaky-vs-real classification.
- Predictive failure forecasting and risk identification.
- A dashboard for data-driven engineering decisions.

**Test optimization**
- Risk-based test selection and prioritization from code-change diffs.
- Coverage-gap and redundancy detection.
- Runtime reduction that **preserves a stated confidence bound**.

**Automation**
- LLM-assisted test generation from requirements and code changes.
- Adaptive / self-healing execution, gated behind a verdict layer.

## The distinctive idea

Most test-optimization tools give you a speedup and ask you to trust it.
`assaylab` emits a **signed receipt that bounds the confidence lost** when it
reduces a suite: *ran subset S → probability of missing regression-class C ≤ ε*,
with an attested, checkable proof. Speedup **with** a formal confidence claim.

Verdicts follow the `agentsensory` contract (Report = verdict + grounded issues
+ Handoff), so results are portable and auditable.

MIT © 2026 Amit Patole

*— amitpatole*
