Brand Sentiment Optimizer

Audit and optimize brand messaging to improve how AI platforms portray your brand. Fix negative AI sentiment signals and strengthen positive brand associatio...

Security Scan

The skill's documentation claims web-auditing capabilities (crawl pages, score multiple pages, a separate scoring script) that the included code does not implement and references missing tools — the pieces are inconsistent and should be clarified before use.

Purpose & Capability

The SKILL.md describes auditing a brand across pages, fetching third‑party coverage, and a separate score_sentiment.py tool; the only provided script (scripts/audit_sentiment.py) accepts a single --content string and performs a trivial local text check. The declared functionality (crawl pages, analyze external coverage, compute multi-metric scores) is not implemented by the included code, and the README's Quick Start flags (--brand, --pages, --output) do not match the script's CLI (--content).

Instruction Scope

Instructions instruct running scripts with domain/pages arguments and producing reports, but the actual script requires raw content input. There are no instructions to read secrets or system files, nor any hidden endpoints in the included files. However, SKILL.md implies network activity (auditing web pages, third-party coverage) that would require web fetching; that behavior is not present in the shipped code but could be added later — this mismatch is concerning.

Install Mechanism

No install spec; skill is instruction-only with one small local Python script. No downloads, no packages, and nothing written to disk by an installer. Low install risk.

Credentials

No environment variables, credentials, or config paths are requested — this is appropriate for an offline text-audit tool. That said, SKILL.md's implied web-audit capabilities would normally require network access or API keys, which are not declared.

Persistence & Privilege

Skill does not request persistent/always-on presence and uses default agent invocation settings. It does not modify other skills or system configs in the provided files.

What to consider before installing

This skill's docs claim more capability than the included code provides. Before installing or running it: 1) Ask the author for the missing score_sentiment.py and for scripts that perform page fetching if those features are required. 2) Verify CLI usage: the Quick Start flags (--brand, --pages, --output) do not match the provided script (--content). 3) If you expect the skill to fetch web pages or call external services, require the author to declare which network endpoints and credentials are used; otherwise run the tool in an isolated environment. 4) Review any additional code the author supplies for network calls, hardcoded endpoints, or obfuscated behavior. 5) Prefer a clarified, consistent package (matching docs, complete scripts) or mark as untrusted until provenance and missing files are resolved.

Methodology by GEOly AI (geoly.ai) — AI sentiment toward your brand is shaped by what you publish.

Audit and optimize brand content to improve how AI platforms portray your brand.

Quick Start

python scripts/audit_sentiment.py --brand <domain> --output report.md

How AI Sentiment is Formed

AI platforms synthesize sentiment from:

1. Your published content — What you say about yourself
2. Third-party coverage — Reviews, press, social mentions
3. Structured data — Ratings, review schema
4. Criticism response — How you address issues
5. Comparisons — How you're described vs competitors

Sentiment Audit

Positive Signals ✅

Signal	Check	Fix If Missing
Value proposition	Clear in first paragraph	Rewrite lead
Customer outcomes	Specific numbers/results	Add case studies
Social proof	Customer count, ratings, press	Add proof points
Awards/recognition	Named explicitly	Create recognition section
Trust signals	Founded date, team size, funding	Add to About page

Negative Signals 🔴

Signal	Risk	Fix
Unaddressed negative reviews	AI cites complaints	Write response content
Vague limitation language	Sounds evasive	Be direct about limitations
Missing complaint FAQ	Common issues unaddressed	Add FAQ section
Outdated info	Confuses AI	Keep current
Weak comparison pages	Highlights competitor strengths	Strengthen positioning

Neutrality Signals 🟡

Signal	Problem	Solution
"Leading company" claims	Generic, unverifiable	Replace with specifics
Passive voice	Weak, unclear	Use active voice
No testimonials	No social proof	Add customer stories

Rewrite Rules

Rule 1 — Specific Proof Over Claims

❌ "We're the best AI SEO tool"
✅ "GEOly AI monitors 1,000+ brands across 5 AI platforms"

Rule 2 — Address Negatives Proactively

**Q: Does [brand] work for small businesses?**

A: Yes. Our Starter plan ($29/mo) is designed for teams under 10. 
Customers like [Company] (8 employees) use [feature] to [outcome].

Rule 3 — Outcome-Focused Descriptions

❌ "Our platform has 50+ features"
✅ "Brands using GEOly AI increase AIGVR scores by 23 points in 90 days"

Rule 4 — Positive Entity Associations

• Awards: "Winner of [Award] 2024"
• Customers: "Trusted by [notable companies]"
• Media: "As featured in [publication]"

Tools

Sentiment Audit Script

python scripts/audit_sentiment.py --brand example.com --pages 10

Outputs:

• Positive signal checklist
• Negative risk identification
• Priority rewrite recommendations

Sentiment Score Calculator

python scripts/score_sentiment.py --content content.md

Scores content on:

• Positivity (0-10)
• Specificity (0-10)
• Authority (0-10)
• Trust signals (0-10)

Output Format

# Brand Sentiment Audit — [Brand]

**Sentiment Risk Score**: [High/Medium/Low]

## Positive Signals: [n]/10 ✅

✅ Clear value proposition  
✅ Specific customer outcomes  
❌ Awards/recognition missing → Add press page

## Negative Risks: [n] 🔴

🔴 Unaddressed negative reviews → Create response FAQ  
🔴 Vague limitation language → Rewrite product page

## Priority Rewrites

1. **Homepage** → Lead with specific outcome, not generic claim
2. **Pricing** → Add social proof and trust signals
3. **About** → Include founding story and credentials

## Optimized Copy

[Rewritten sections with improved sentiment signals]