# Unknowns Reference Map

> Prove semantic understanding before porting, reimplementing, or matching a reference implementation. Use when the user points at another file/package/language/repo as the desired behavior, asks for a semantics map, or wants confirmation before Codex writes code based on a reference.

- Skill: `infinite-labs-ai/unknowns-reference-map` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add infinite-labs-ai/unknowns-reference-map`
- Raw SKILL.md: https://api.skillmd.com/api/skills/infinite-labs-ai/unknowns-reference-map/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: Infinite-Labs-AI (https://skillmd.com/u/infinite-labs-ai)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/infinite-labs-ai/unknowns-reference-map

---


# Unknowns Reference Map

## Workflow

Map behavior before porting.

Return:

- Reference inventory: key behaviors, invariants, edge cases, and tests.
- Source-to-target semantic pairs.
- Preserved behavior, adapted behavior, deliberately dropped behavior.
- Gotchas where syntax translation would break semantics.
- Tests required to prove equivalence.
- Confirmation/correction prompt.
- Stop condition: the user confirms the map or corrects it before implementation starts.

## Rules

Do not implement until the semantics map is accepted unless the user explicitly overrides that gate. Distinguish exact equivalence from intentional product/API adaptation.

Use HTML when side-by-side code excerpts, highlighted notes, and correction buttons make review easier.