What is deterministic sampling in this context?

It is a strategy for selecting a consistent subset of tests that provide maximum coverage for the least amount of time, ensuring agents have a reliable and fast validation gate.

How does this skill improve agent productivity?

It creates tiered feedback loops and fast 'smoke test' subsets, allowing agents to catch obvious regressions in seconds rather than waiting for long-running full test suites.

What is an oracle in autonomous development?

An oracle is a source of truth, such as a reference implementation, a legacy system, or golden files, used to programmatically verify that an agent's changes are functionally correct.

Why is structured error output important for AI?

Standardized output allows AI agents to parse failures programmatically, helping them understand exactly what was expected versus what happened so they can self-correct without human intervention.

Can I use this to run my existing tests?

No, this skill is for designing and structuring your testing strategy for autonomy. You should continue to use your project's native test runner (like Vitest, Pytest, or Jest) to execute the actual commands.

Test-Driven Autonomous Development

Name: Test-Driven Autonomous Development
Author: am2rican5

byam2rican5

0•

安全与测试

Establishes robust test suites and feedback loops that allow AI agents to work autonomously with high confidence and minimal human intervention.

This skill provides a comprehensive framework for designing 'agent-ready' testing environments, specifically optimized for Claude-driven workflows. It guides users through creating reliable oracles—sources of truth such as reference implementations or property-based tests—and organizing them into efficient tiers. By implementing deterministic sampling and structured error reporting, the skill ensures that autonomous agents receive the rapid, actionable feedback they need to verify their own code changes, making it ideal for complex migrations, refactors, and system-level development tasks.

主要功能

01Tiered test design for balancing rapid feedback with comprehensive system validation

020 GitHub stars

03Deterministic sampling techniques to create high-speed 'smoke test' subsets

04Baseline measurement protocols to establish correctness before autonomous execution

05Oracle-based testing strategies including reference implementation and snapshot verification

06Machine-readable error output formatting optimized for agent interpretation

使用场景

01Creating high-confidence feedback loops for large-scale behavioral parity testing

02Setting up autonomous agent workflows for compiler or specialized language development

03Designing a verification harness for legacy system migrations or API refactoring

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add am2rican5/cc-plugin test-driven-autonomous-dev

For use in Claude.ai and ChatGPT

主要功能

01Tiered test design for balancing rapid feedback with comprehensive system validation

020 GitHub stars

03Deterministic sampling techniques to create high-speed 'smoke test' subsets

04Baseline measurement protocols to establish correctness before autonomous execution

05Oracle-based testing strategies including reference implementation and snapshot verification

06Machine-readable error output formatting optimized for agent interpretation

使用场景

01Creating high-confidence feedback loops for large-scale behavioral parity testing

02Setting up autonomous agent workflows for compiler or specialized language development

03Designing a verification harness for legacy system migrations or API refactoring