greentechgear/citation-labs-os-skill

This repository packages the Citation Labs Operating System by Garrett French and Citation Labs as an Agent Skill for Claude Code, Claude.ai, and any other agent that reads SKILL.md.
The OS is Citation Labs' methodology for finding where a brand is not showing up in AI answers, what information is missing, what needs to change, and whether the change worked. It is built from four field guides and six assistant task modules, held together by a strict discipline about evidence: every finding keeps the strength it actually earned, and nothing gets upgraded on its way to the next step.
The official way to use it is to download the four guides, upload them to a chat, and paste a starter prompt. This skill does that setup once, permanently. Claude loads the router and rules when the topic comes up, and pulls in only the guide sections and task instructions the current question needs.
## What it does
Ask in plain language. The skill routes the request to one of the OS's six independent tasks and follows the governing guide.
| You ask | Task | You get back |
| "What do ChatGPT and Perplexity say and cite about us for these prompts?" | 01 Establish a baseline | AI Visibility Plan + Readout with coverage, the six visibility zones checked separately, and a diagnosis |
| "Our CMO, CFO, and content lead all weigh in on this. What does each need?" | 02 Shared decision | Decision Evidence Record with per-responsibility findings |
| "Turn these query fan-outs and uncited Reddit threads into interview questions" | 03 Interview questions | Passage-grounded question queue with respondent criteria |
| "Make something useful out of these interview transcripts" | 04 Prepare information | Production brief or supported draft with a claim-support table and review list |
| "We published a security evidence page. Did it work?" | 05 Plan or evaluate a test | Measurement contract or readout separating AI effects from human usefulness |
| "Did this visibility win actually matter to the business?" | 06 Business consequence | Business Impact Readout at the supported evidence strength, or a precise inquiry plan |
Say "walk me through the Citation Labs OS" with no specific problem and it runs Citation Labs' own starter flow: a short welcome, an ASCII map of the guides, and four tailored paths (Executive Tour, Diagnose a Problem, Test a Customer Scenario, Stress-Test the Framework).
It works without any paid tools. If you have the Citation Labs instruments (Thinking Inspector or XOFU) connected as MCP servers, the skill knows what each can and cannot establish and asks before anything that costs money.
## Install
### Claude Code (recommended)
/plugin marketplace add greentechgear/citation-labs-os-skill
/plugin install citation-labs-os@citation-labs-os
Or copy the skill folder in by hand:
bash
git clone https://github.com/greentechgear/citation-labs-os-skill
cp -r citation-labs-os-skill/skills/citation-labs-os ~/.claude/skills/
### Claude.ai and Claude Desktop