compliance-checker

Files (3)

• LICENSE.txt
• SKILL.md
• openai.yaml · agents/

SKILL.md (first lines)

# Compliance Checker

Audit affiliate content for FTC compliance, platform-specific rules, and legal requirements. Checks disclosure placement, prohibited claims, endorsement guidelines, and platform policies. Output is a compliance scorecard with issues, severity, and fix suggestions.

## Stage

S8: Meta — The FTC has fined affiliates $4.2M+ for undisclosed endorsements. One missing disclosure can result in legal action, platform bans, or program termination. This skill is the safety net — run it on any content before publishing to catch compliance issues before they become problems.

## When to Use

- User wants to check content before publishing
- User asks about FTC rules or affiliate disclosure requirements
- User is unsure if their content is compliant
- User says "is this legal?", "do I need a disclosure?", "check my post"
- User is preparing content for a platform with strict ad policies (Facebook, Google)
- Chaining: run after any S2-S5 or S7 content-producing skill before publishing
- User wants to audit existing published content

## Input Schema

```yaml
content: string                # REQUIRED — the content to check (text, markdown, or HTML)

content_type: string           # REQUIRED — "social_post" | "blog" | "landing_page"
                               # | "email" | "ad" | "video_script"

platform: string               # OPTIONAL — "linkedin" | "twitter" | "reddit" | "facebook"
                               # | "tiktok" | "youtube" | "google_ads" | "pinterest"
                               # Platform-specific rules applied if provided

claims:                        # OPTIONAL — specific claims to verify
  - string                     # e.g., ["earn $10K/month", "guaranteed results"]
```

**Chaining context**: If content was produced by S2-S5 or S7 in the same conversation, pull it directly. The user should not have to paste content that was just generated.

## Workflow

### Step 1: Detect Affiliate Links

Scan content for:
- URLs with affiliate parameters (`ref=`, `aff=`, `partner=`, UTM tags)
- Shortened URLs (bit.ly, etc.) that may hide affiliate links
- Product mentions that imply a commercial relationship

### Step 2: Check FTC Disclosure

Read `shared/references/ftc-compliance.md` for rules. Check:
- **Presence**: Is there a disclosure? (required if any affiliate link exists)
- **Placement**: Is the disclosure before or near the affiliate link? (not buried at the bottom)
- **Clarity**: Is it clear to a reasonable consumer? ("affiliate link" is clear; "partner" alone is not)
- **Format by content type**:
  - Social post: `#ad` or `Affiliate link` visible without expanding
  - Blog: Disclosure in the opening paragraph, above the fold
  - Landing page: Medium disclosure above the fold
  - Email: Disclosure near the affiliate link
  - Ad: Platform-specific requirements

### Step 3: Check Prohibited Claims

Scan for:
- **Income claims**: "earn $X", "make money fast", "passive income guaranteed"
- **False urgency**: "only 3 left" (if not verifiable), "offer expires" (if no real deadline)
- **Health/medical claims**: unsubstantiated health benefits
- **Guaranteed results**: "guaranteed to work", "100% success rate"
- **Fake scarcity**: "limited spots" (if not actually limited)
- **Fake testimonials**: results that aren't typical without disclaimer

### Step 4: Check Platform Rules

If `platform` is provided, apply platform-specific rules:
- **Reddit**: Self-promotion rules (10:1 ratio), must disclose in post
- **Facebook/Instagram**: Branded Content tool, "Paid Partnership" label for ads
- **Google Ads**: Clear commercial intent, no misleading claims, landing page requirements
- **TikTok**: #ad or Paid Partnership toggle, no medical/financial advice claims
- **YouTube**: Verbal + written disclosure in first 30 seconds, "Includes paid promotion" checkbox

### Step 5: Score and Report

Rate compliance on three levels:
- **PASS**: No issues found
- **WARN**: Minor issues that should be fixed (e.g., disclosure placement could be better)
- **FAIL**: Critical issues that must be fixed before publishing (e.g., no disclosure at all)

### Step 6: Generate Fixes

For each issue, provide:
- What's wrong (specific quote from content)
- Why it matters (rule reference)
- How to fix it (specific replacement text)

### Step 7: Self-Validation

Before presenting output, verify:

- [ ] All affiliate links in the content are detected and flagged
- [ ] Disclosure placement check matches platform-specific rules
- [ ] Prohibited claims identified with exact quotes from content
- [ ] Fix suggestions are copy-paste ready and preserve original tone
- [ ] Corrected content would pass a re-scan by this same skill

If any check fails, fix the output before delivering. Do not flag the checklist to the user — just ensure the output passes.

## Output Schema

```yaml
output_schema_version: "1.0.0"  # Semver — bump major on breaking changes
compliance:
  overall_score: string        # "PASS" | "WARN" | "FAIL"
  disclosure_present: boolean
  disclosure_placement: string # "correct" | "needs_improvement" | "missing"
  prohibited_claims: number    # count of issues found
  platform_issues: number      # count of platform-specific issues

issues:
  - severity: string          # "critical" | "warning" | "info"
    category: string          # "disclosure" | "claims" | "platform" | "formatting"
    description: string       # what's wrong
    quote: string             # the problematic text

From Affitor/affiliate-skills, MIT.

More from Affitor/affiliate-skills