agentsop-regression-gate

Community

Block LLM regressions with CI-eval gates.

Authoragentsope
Version1.0.0
Installs0

System Documentation

What problem does it solve?

This Skill prevents LLM prompt, model, and retriever changes from silently shipping behavior regressions by enforcing a held-out evaluation gate in CI.

Core Features & Use Cases

  • Held-out eval gate: builds (then freezes) a golden eval set, runs it on every change, and blocks merges on score drops.
  • Metric + threshold enforcement: consumes a calibrated metric and applies an absolute floor and/or relative no-regression delta above measured noise.
  • Cross-framework CI wiring: assembles the discipline across common eval stacks (e.g., LlamaIndex, DSPy, promptfoo, LangSmith) so teams can implement a consistent regression gate.

Quick Start

Activate regression-gate for every prompt/model/retriever change by asking the AI to design your eval set, select a calibrated metric, set a noise-aware threshold, and produce a CI wiring plan that fails the build on regressions.

Dependency Matrix

Required Modules

None required

Components

references

💻 Claude Code Installation

Recommended: Let Claude install automatically. Simply copy and paste the text below to Claude Code.

Please help me install this Skill:
Name: agentsop-regression-gate
Download link: https://github.com/agentsope/SkillAlchemy/archive/main.zip#agentsop-regression-gate

Please download this .zip file, extract it, and install it in the .claude/skills/ directory.
View Source Repository

Agent Skills Search Helper

Install a tiny helper to your Agent, search and equip skill from 471,000+ vetted skills library on demand.