Home · Skills · Variance Narrative Generator
Back to all skills

Variance Narrative Generator

Generates ownership-ready variance narratives from budget-vs-actual reports. Screens for materiality, classifies variances as timing/permanent/one-time/trend, projects full-year NOI impact, and drafts investor-quality explanations.

What this skill does

Variance Narrative Generator is an A.CRE Intelligence Hub skill that gives your AI agent the analyst-grade workflow a senior commercial real estate professional would run. Generates ownership-ready variance narratives from budget-vs-actual reports. Screens for materiality, classifies variances as timing/permanent/one-time/trend, projects full-year NOI impact, and drafts investor-quality explanations. Generates ownership-ready variance narratives from budget-vs-actual reports. Screens for materiality, classifies variances as timing/permanent/one-time/trend, projects full-year NOI impact, and drafts investor-quality explanations. Activate it inside ChatGPT, Claude, Manus, or OpenClaw by saying something like "Use the Variance Narrative Generator skill" — the Hub routes the request to this skill and the underlying primary-source CRE data feeds backends automatically. Built and maintained by Mario Urquia, with live primary-source data so the numbers your AI returns are the numbers a real CRE analyst would use.

How you'll use it

"Use the Variance Narrative Generator skill"

Say something like this to your AI agent in the Hub, Claude, or ChatGPT to activate this skill.

Skill activation rules

Detailed routing logic, prompts the skill responds to, and operational guardrails as documented by the author.

Generates ownership-ready variance narratives from budget-vs-actual reports. Screens for materiality, classifies variances as timing/permanent/one-time/trend, projects full-year NOI impact, and drafts investor-quality explanations.

What's inside

Variance Narrative Generator

You are a variance narrative engine for CRE property reporting. Given a budget-vs-actual report, you screen for materiality, classify each variance (timing, permanent, one-time, trend), project full-year NOI impact, and draft ownership-ready…

Show full previewShow less

Variance Narrative Generator

You are a variance narrative engine for CRE property reporting. Given a budget-vs-actual report, you screen for materiality, classify each variance (timing, permanent, one-time, trend), project full-year NOI impact, and draft ownership-ready narratives. You turn a 20-40 minute manual write-up per property into a reviewed-and-ready first draft. Your language is professional, factual, and action-oriented -- no hedging, no vague qualifiers. These narratives go to property owners and institutional investors.

When to Activate

Trigger on any of these signals:

  • Explicit: "write variance narrative", "explain budget variances", "variance report for [property]", "what drove the NOI miss"
  • Implicit: user provides a budget vs. actual report; user asks why expenses are over budget; user mentions monthly close reporting
  • Cycle-driven: monthly close, quarterly investor reporting, annual review

Full instructions and any reference files ship in the .skill bundle.

Bundle structure
variance-narrative-generator/
├── SKILL.md                     # Required: instructions + metadata
└── references/                  # Optional: documentation
    └── narrative-patterns.yaml  # 20.8 KB

Who built it

With contributions from Avi Hacker.

How to run it

Not in the Hub? Download the .skill bundle above and follow the A.CRE skills install guide → to load it into Claude Desktop, ChatGPT, Cursor, or any other agent that supports skills.

Already in the Hub. If you're an AI.Edge Pro or A.CRE Accelerator member, this skill is bundled into the A.CRE Intelligence Hub direct connector (MCP server) — just ask your agent.

About this skill

Third-Party skill. This skill is redistributed under the Apache-2.0 license and is not authored or endorsed by Adventures in CRE. Use is governed by the upstream license and the original creator's terms. Original attribution: Mario Urquia (link) and contributors (Avi Hacker).