icfpc-problem-notes

Reads and updates per-problem knowledge notes for ICFPC 2026 Littleman contest work.

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

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When resuming work on an ICFP Contest problem, it is easy to repeat failed searches, lose verified findings, or treat stale scores as current. This Skill centralizes per-problem knowledge notes so confirmed architectures, invariants, rejected search families, and restart points are read before work and updated after work. ## Core Features & Use Cases - Pre-work review: Resolves the target problem via the problem catalog and reads its note in member-18/docs/problems/ to surface known constraints, failed approaches, and restart points before starting. - Evidence-backed updates: Records only verified findings such as validated architectures, invariants, counterexamples, exhaustive search bounds, and UNSAT results, with dates, suite fingerprints, and source SHAs. - Telemetry and verification: Tracks runs with the skill-telemetry tool and validates note changes with the problem-notes test suite and git diff checks. - Use Case: Before optimizing a Littleman solution for a specific problem, ask what is already known; the Skill reports confirmed facts, bounded negative results, open questions, and where to resume, then records new durable findings afterward. ## Quick Start Use the icfpc-problem-notes skill to review what is already known about the Reverse a List problem and record any new verified findings from this session.

Frequently Asked Questions about icfpc-problem-notes

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

FAQPage Schema
How do I check what is already known about an ICFPC problem before working on it?▼

Resolve the problem by ID, slug, or display name using the problem-catalog.mjs tool, then read its note in member-18/docs/problems/. The note separates confirmed facts, bounded negative results, open questions, and restart points.

How do I record experiment results in the problem notes?▼

Update the note only when you have verified findings such as validated architectures, invariants, reproducible counterexamples, or exhaustive search bounds. Follow the _template.md section structure, update the date, and run the problem-notes test suite afterward.

Does this skill judge or submit Littleman solutions?▼

No. Judging, best-solution management, and official submission are delegated to the icfpc-local-judge skill. This skill only reads and maintains knowledge notes and never registers .man artifacts itself.

What kinds of findings should not be added to the notes?▼

Do not promote mere ideas, unreproduced speculation, or in-progress states to confirmed knowledge. Only verified improvements, invariants, counterexamples, exhaustive search results, and durable restart points qualify for inclusion.

Why should old scores in the notes not be treated as current?▼

Notes record scores as snapshots with acquisition date, suite fingerprint, source SHA, and submission ID. Team bests are mutable, so scores from different suite fingerprints must not be compared directly.