dim-s

Prompt Atlas — AI skill for Claude Code

AI community

Auditable, citation-driven LLM prompt review skill across 9 frontier vendors and 2-9B small-local models.

How to install Prompt Atlas

This entry records only its repository, not the path inside it, so there is no exact command to give. Open dim-s/prompt-atlas and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Prompt Atlas does

Auditable, citation-driven LLM prompt review skill across 9 frontier vendors and 2-9B small-local models.

Alternatives in AI

Pilotfish — Multi-model orchestration layer for Claude Code — the frontier model plans, cheaper models execute, verificati 688 ★
Maestro Companion — Quick execution for small tasks — minimal run lifecycle (start + done) with evidence recording 530 ★
Unofficial Claude Code Prompt Playbook — Unofficial playbook for production-grade LLM system prompt architecture, derived from local analysis of Claude 258 ★

README

prompt-atlas

An auditable, citation-driven skill for **reviewing and improving the wording of LLM prompts** across all major current model families. Every recommendation cites a documented behavior (model × axis cell) so the user can challenge it from data, not authority.

License: [CC-BY-SA 4.0](./LICENSE). Built on the methodology of [matrix-citation prompt review](references/matrix.md).

Coverage

Class 1 — Frontier agent-facing prompts (11 vendor families)

Vendor	Models covered
Anthropic Claude	Fable 5 (Jun 2026, tier above Opus) / Opus 5 (Jul 2026) / Opus 4.8 / 4.7 / Sonnet 5 (Jun 2026) / Sonnet 4.6 / Haiku 4.5 + legacy 4.6
OpenAI GPT-5.x in Codex CLI	GPT-5.6 Sol / Terra / Luna (Jul 2026; August update: Chat = August model, Codex/Work stay July) / 5.5 (+ Instant) / 5.4 / 5.3 / 5.3-codex / 5.2 / 5.1
Google Gemini 3.x in Gemini CLI	3.1 Pro / 3.7 Flash (Aug 2026) / 3.6 Flash (Jul 2026) / 3.5 Flash / 3 Flash / 3.5 Flash-Lite (Jul 2026) / 3.1 Flash-Lite + 2.5 legacy
Moonshot Kimi	K3 (Jul 2026) / K2.7-Code (Jun 2026) / K2.6 / K2.5 (K2 retired 2026-05-25)
Z.ai GLM	GLM-5.3-Flash (Aug 2026, multimodal cost sibling, a.k.a. ox-alpha) / GLM-5.3 (Aug 2026) / GLM-5.2 / GLM-5.1 / GLM-5 / GLM-4.6
Alibaba Qwen frontier	Qwen3.8-Max (GA Aug 2026, open weights 2.4T-A95B, QwenCloud) / Qwen3.8-Flash (Aug 2026, 6B active, open `Qwen3.8-Flash-Next`) / Qwen3.7-Max / 3.7 Plus / 3.6 Plus / 3.6 Max-Preview / 3-Max-Thinking
DeepSeek	V4-Pro (GA update 2026-08-13) / V4-Flash / V4-Flash-Vision-Exp (Aug 21 2026, multimodal) / V3.2 / R1 (legacy `deepseek-chat` / `deepseek-reasoner` aliases retired 2026-07-24)
xAI Grok	Grok 4.6 (Aug 2026, 500K context) / Grok 4.5 (Jul 2026, 500K) / Grok 4.3 (1M context) — Grok 5 still unshipped
Mistral frontier	Mistral Large 3 / Mistral Medium 3.5 (Apr 2026) / Mistral Small 4 /