icfpc-local-judge

Judge ICFPC Littleman solutions with the official WASM engine and manage submissions.

1|Updated Jul 30, 2026
One-click install
npx skills add https://github.com/d-experts/public-icfpc-2026 --skill icfpc-local-judge-d-experts
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: icfpc-local-judge
Source: https://github.com/d-experts/public-icfpc-2026/tree/main/member-18/.agents/skills/icfpc-local-judge
Command: npx skills add https://github.com/d-experts/public-icfpc-2026 --skill icfpc-local-judge-d-experts

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? It provides a consistent, safe workflow for scoring ICFPC 2026 Littleman .man solutions locally with the official WASM judge engine, tracking content-addressed best solutions, and managing official submissions through the team proxy without reimplementing scoring or persistence logic. ## Core Features & Use Cases - Local Judging with WASM Engine: Runs the repository's judge.mjs CLI with LITTLEMAN_JUDGE_ENGINE=wasm to score solutions against the public test suite, avoiding known inaccuracies of the native runner. - Best-Solution Store & History: Maintains a content-addressed store of accepted bests, lists current bests, inspects saved sources, and shows accepted-revision history per problem. - Official Submission Management: Auto-submits newly created or improved bests through the team proxy with idempotency keys, resumes interrupted submissions, and reports official scores and attempt history. - Live HTML Viewer: Serves a localhost viewer showing current bests, histories, official submissions, and source text, refreshed on each page load. - Use Case: After writing an improved solution for problem 1-3, ask the assistant to judge it; the skill scores it locally, saves it as the new best if improved, auto-submits it officially, and reports both local and official results. ## Quick Start Ask the assistant to judge your solution file for a problem such as 1-3 using the local judge, update the saved best, and submit it officially if it improves.

Frequently Asked Questions about icfpc-local-judge

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I judge a Littleman .man solution locally?▼

Run the repository's judge.mjs CLI with the problem ID or slug and the solution file, always setting LITTLEMAN_JUDGE_ENGINE=wasm. The command scores the solution against the public suite and reports the exact local score and store status.

Why must the WASM engine be used instead of the native runner?▼

The Rust native runner is faster but has produced incorrect judging results, and the CLI's bare default selects it when preparation succeeds. The skill forces LITTLEMAN_JUDGE_ENGINE=wasm on every judge invocation unless native debugging is explicitly requested.

How does automatic official submission work after judging?▼

The default judge command auto-submits only a newly saved created or strictly improved local best through the team proxy. Same, tied, worse, failed, or --no-save results are never auto-submitted, and the content-addressed saved source is what gets submitted.

Can I compare local public scores with official submission scores?▼

No, local public scores and official submission scores cover different case sets and must be kept separate. The official score comes from the upstreamResponse in the proxy's individual submission GET response.

What happens if a submission is interrupted before completion?▼

The store acts as a durable queue: it records an attempt UUID before POST and uses it as the proxy Idempotency-Key. The CLI can resume polling with the known submission ID or safely repeat the same payload with the same attempt UUID.

How do I view current bests and submission history in a browser?▼

Run the judge CLI's viewer command to start a localhost server on 127.0.0.1 that fetches current bests, histories, submissions, and source text on every page load. Use --open to launch it automatically or --snapshot for a portable static HTML file.