Evaluate specification file quality with an LLM-as-judge scorecard.
npx skills add https://github.com/DMokong/speculator --skill spec-score
Please help me install this Agent Skill. Skill: spec-score Source: https://github.com/DMokong/speculator/tree/main/skills/spec-score Command: npx skills add https://github.com/DMokong/speculator --skill spec-score
This Skill automates the evaluation of specification quality, ensuring that development requirements are clear, complete, and meet predefined standards before implementation begins.
Run the spec-scorer agent to evaluate the quality of the specification file.
High-intent search queries and answers about installing and using this skill.
To evaluate specification quality, you can use an LLM-as-judge agent to automatically score a spec file against quality standards. This generates a detailed scorecard that identifies specific areas for improvement, reducing rework before implementation begins.
An LLM-as-judge agent for development workflows is an automated evaluator that scores requirement specifications for completeness and clarity. It acts as a quality gate to prevent ambiguous or incomplete specs from proceeding into the coding phase.
You automate spec quality scoring by running a scorer agent that requires a project configuration and the spec file path. The agent evaluates the document and outputs an objective scorecard with actionable feedback and flags.
Yes, you need a project configuration and the spec file path to run a spec quality evaluation. These inputs are required for the LLM-as-judge agent to understand the context and generate an accurate quality scorecard.
You should enforce a specification quality gate at the initial stage of an agentic development workflow. Applying this gate before coding ensures that requirement specs meet predefined completeness standards and prevents low-quality inputs.
Obsidian-OwlInitialize Specwright by detecting the development stack and generating configuration files.
xhyqaqTransform rough product ideas into approved design specifications through guided dialogue.
GobbyAIEvaluate AI agent performance using multi-dimensional rubrics and LLM-as-judge.
y-miyazakiReviews SKILL.md files for structural requirements, quality standards, and design patterns.
