Score a draft specification against rubric_notes.md with console-only output.
npx skills add https://github.com/Jason5330/ai-self-eval --skill cheat-score
Please help me install this Agent Skill. Skill: cheat-score Source: https://github.com/Jason5330/ai-self-eval/tree/main/skills/cheat-score Command: npx skills add https://github.com/Jason5330/ai-self-eval --skill cheat-score
Scores a single draft specification, learning topic, or coding spec by applying a rubric, without performing any predictions. It outputs results to the console only, allowing quick, non-destructive evaluation to decide whether to proceed to the formal predict phase.
Provide a draft path and rubric_notes.md to the cheat-score workflow to receive a non-predictive score, a composite, and a recommended next step.
High-intent search queries and answers about installing and using this skill.
You can score a draft specification non-predictively by providing a draft path and rubric_notes.md file. The workflow evaluates the draft against the rubric to produce per-dimension scores and a composite, printing results directly to the console.
Non-predictive rubric assessment evaluates a draft specification against scoring criteria without generating downstream predictions. You need it to quickly decide whether a draft is ready to proceed to a formal predict phase without triggering expensive resource consumption.
Console-only rubric scoring does not write any files to your workspace. It performs a non-destructive evaluation that prints per-dimension scores, a composite, and a recommended next step exclusively to the console output for quick auditing.
To evaluate a coding spec against custom rubric notes, provide the draft-path and rubric_notes.md to the workflow. The evaluation is routed through a blind sub-agent that applies the rubric criteria and outputs per-dimension scores to the console.
The limitations of a non-predictive draft assessment workflow include lacking automated file generation and being unable to produce downstream predictions. It outputs only console text, requiring manual review before advancing to the formal predict phase.
Jason5330Automate rubric and bucket boundary upgrades with auditable governance.
nahisahoQuantify machine learning predictive uncertainty with conformal prediction, MC dropout, and deep ensembles.
alexmarucciPlan, generate, and evaluate code through a multi-agent pipeline.
XBuilderLABScore scripts blindly using rubric notes without user context or past data.
