CLEA Core

Prompts are the questions and instructions CLEA uses to collect or act on evidence. They are the basic unit of work.

Some prompts measure visibility. Other prompts tell agents what to do after data exists.

Mention prompts are buyer-style questions that check whether your brand appears in AI answers.

They also reveal competitors and sources. This makes them one of the most important parts of the workspace.

Agent prompts tell CLEA to inspect data, use files, and produce an output. The output can be a suggestion card, report, summary, or draft.

Agent prompts are stronger when they reference real workspace evidence.

Evidence objects are the things CLEA can point to when it explains a recommendation. They include chatlogs, sources, citations, prompt results, and connected analytics.

Evidence keeps the workspace honest.

Chatlogs are saved records of prompt answers and CLEA interactions. They make it possible to inspect what was asked, what was answered, and what changed later.

Sources are pages or domains that AI systems read or cite while forming an answer.

They are useful because they show what the answer engine may trust.

Citations are source references connected to an answer. They can help explain why a competitor was named or why a claim appeared.

Visibility analytics turns prompt evidence into dashboard views. It helps you compare mentions, competitors, sources, sentiment, and movement over time.

Suggestion cards turn evidence into action. A good card gives one useful next step and a CTA.

Cards can come from onboarding, scheduled agents, manual chat, or analytics gaps.

Sitemap expansion connects visibility gaps to website work. It is strongest when prompt and source evidence already show what page should exist or improve.

For GitHub-backed expansion, CLEA expects the /clea-sitemap-expansion folder in the connected repository root.

Scheduled workflows give CLEA a rhythm. They can review prompt runs, inspect files, prepare cards, and support recurring reports.

Use them after the first manual review is clear. Automation should follow a proven routine.