Content Calendar Skill#

Generate a full month of social content in minutes. Paste in your niche and audience — get back a structured, ready-to-schedule calendar with varied hooks, body copy, CTAs, and hashtags.

Free vs Premium#

Free tier (no API key needed):

• --demo — generates a complete 7-day real estate sample calendar, zero API calls
• --version — print skill version
• --compliance-only — check any copy for forbidden words before posting

Free tier (API key required):

• 7-day calendar on any platform/niche
• Structured JSON + human-readable output
• Single platform per run

Premium features (same API key, no upgrade needed):

• --days=30 — full 30-day calendars
• --platforms=all — LinkedIn + X + Facebook + Instagram in one run
• --ab-variants — 2 hook variants per post for A/B testing
• --csv — export to CSV for Buffer, Hootsuite, Notion, or any scheduler
• --theme — weekly theme override (e.g., "Week 2: objection busting only")
• --min-score=75 — quality gate: posts scoring below threshold are auto-rewritten (max 2 passes)
• --past-results=results.json — feed in engagement data from previous runs; skill adapts theme weights toward what worked
• --tracker — output blank engagement tracker template alongside the calendar (paste into Notion/Sheets)

What this skill does#

Generates structured content calendars for any niche with:

• Variety enforcement — topics rotate so nothing repeats within 7 days
• Platform-aware formatting — character limits, hashtag counts, and tone matched per platform
• Content mix — market education, social proof, objection busting, community spotlight, CTAs (5-pillar rotation)
• Hook-first structure — every post opens with a pattern interrupt, not a topic statement
• JSON + human-readable output — paste JSON into your scheduler; read the human format to review quickly
• Post-level quality scoring — every post scored 0–100 across 4 dimensions before output
• Auto-rewrite loop — posts below --min-score threshold regenerated with fix instructions (max 2 passes)
• Adaptive theme weighting — pass in past engagement data and the skill shifts content mix toward what converted

Supported verticals#

NicheExample brand voiceReal estateMarket urgency, local expertise, lifestyleMortgage / lendingEducation-first, trust-building, compliance-safeCrypto / DeFiData-driven, opportunity framing, risk-awareSaaS / techProblem-solution, feature → benefit, ROI focusE-commerceProduct-led, social proof heavy, seasonal hooksCoaching / consultingAuthority positioning, transformation storiesHealthcare / wellnessEmpathy-first, outcome-focused, evidence-backedAny nichePass --niche="your niche" and --audience="your audience"

Input contract#

bash

# Minimum — niche and audience
python3 generate_calendar.py --niche="real estate" --audience="first-time buyers"

# Full options
python3 generate_calendar.py \
  --niche="mortgage broker" \
  --audience="move-up buyers" \
  --platform=linkedin \
  --days=30 \
  --tone=educational \
  --ab-variants \
  --csv \
  --output=my_calendar.json

Options:

FlagDefaultDescription--nicherequiredYour brand/industry (e.g. "real estate agent")--audience"general"Target reader (e.g. "first-time homebuyers")--platformlinkedinlinkedin / x / facebook / instagram / all--days77 or 30--toneconversationalconversational / educational / urgent / luxury / bold--ab-variantsoffGenerate 2 hook options per post--csvoffExport calendar as CSV--themeautoWeekly theme override--min-score70Quality gate — posts below this score auto-rewrite (0 to disable)--past-resultsnonePath to engagement JSON from prior run — adapts content mix--trackeroffOutput blank engagement tracker template alongside calendar--demooffRun on built-in sample, zero API calls--versionoffPrint version--formathumanhuman / json

Output contract#

Human-readable output:

text

=== 30-Day LinkedIn Calendar: Mortgage Broker → Move-Up Buyers ===

Week 1 Theme: Market Reality

--- Day 1 (Mon) | LinkedIn | Market Education ---
HOOK: Most move-up buyers don't realize they already have the equity to do this.
BODY: If you bought your home 3-5 years ago, you're sitting on an average of $87,000 in
      equity. That's your down payment for the next place — without touching savings.
      The math most people don't run: sell at today's prices, buy at today's prices,
      and the equity gap covers the difference more than you think.
CTA:  Drop your zip code below — I'll run the numbers for your neighborhood.
HASHTAGS: #RealEstate #HomeEquity #MovingUp #Homeownership
CHARS: 487
SCORE: 81/100 | Hook: 26/30 | Platform: 22/25 | Specificity: 20/25 | CTA: 13/20

JSON output ( --format=json):

json

{
  "metadata": {
    "niche": "Mortgage Broker",
    "audience": "Move-up buyers",
    "platform": "linkedin",
    "days": 30,
    "generated_at": "2026-03-27T17:00:00Z"
  },
  "weeks": [
    {
      "week": 1,
      "theme": "Market Reality",
      "posts": [
        {
          "day": 1,
          "weekday": "Mon",
          "platform": "linkedin",
          "content_type": "market_education",
          "hook": "Most move-up buyers don't realize they already have the equity to do this.",
          "body": "...",
          "cta": "Drop your zip code below — I'll run the numbers for your neighborhood.",
          "hashtags": ["#RealEstate", "#HomeEquity", "#MovingUp", "#Homeownership"],
          "char_count": 487,
          "hook_b": null,
          "engagement_score": {
            "total": 81,
            "hook_strength": 26,
            "platform_fit": 22,
            "specificity": 20,
            "cta_clarity": 13,
            "rewrites": 0
          }
        }
      ]
    }
  ]
}

CSV output ( --csv, compatible with Buffer/Hootsuite/Notion):

text

day,weekday,platform,hook,body,cta,hashtags,char_count
1,Mon,linkedin,"Most move-up buyers...","If you bought your home...","Drop your zip code...","#RealEstate #HomeEquity",487

Post Quality Scoring#

Every generated post is scored across 4 dimensions before output:

DimensionWeightWhat's measuredHook strength30 ptsPattern interrupt quality — does it stop the scroll? Scored on: specificity, unexpected angle, curiosity gapPlatform fit25 ptsCharacter count, hashtag count, tone match, reading level for platformSpecificity25 ptsConcrete numbers, named outcomes, real examples vs. vague generalitiesCTA clarity20 ptsIs the ask clear? Is there exactly one action? Is friction low?

Total: 100 points

Posts scoring below --min-score (default: 70) are automatically rewritten with targeted fix instructions (e.g., "hook is vague — add a specific number or outcome"). Max 2 rewrite passes per post.

Each post in the output includes:

text

SCORE: 81/100 | Hook: 26/30 | Platform: 22/25 | Specificity: 20/25 | CTA: 13/20

Why this matters: A calendar of 30 posts where 8 have weak hooks will underperform a calendar of 30 well-scored posts. The scoring loop catches and fixes low-quality posts before you schedule them.

Adaptive Theme Weighting#

Pass in results from a previous calendar with --past-results=results.json. The skill reads which content types drove the most engagement and shifts the next calendar's mix accordingly.

Tracker format (also generated with --tracker):

json

{
  "period": "March 2026",
  "platform": "linkedin",
  "posts": [
    {
      "day": 1,
      "content_type": "market_education",
      "impressions": 1240,
      "engagement_rate": 0.048,
      "clicks": 22,
      "score_at_generation": 81
    }
  ]
}

What adapts:

• Content mix percentages shift toward top-performing content_type values (e.g., if social_proof drove 3× the engagement, its share increases from 20% → 30% in the next run)
• Weak themes are reduced proportionally
• Platform formatting rules stay fixed (algorithm constraints don't adapt to your data)

Example shift:

text

Past results: social_proof avg engagement 6.2% vs. direct_cta 1.1%
Next run mix: social_proof 30% (+10%), direct_cta 12% (−8%)

Engagement Tracker Template#

Run with --tracker to get a blank tracker alongside the calendar:

text

content-calendar-pro Engagement Tracker
Generated: [date] | Platform: [platform] | Niche: [niche]

Day | Weekday | Type           | Hook (first 60 chars)     | Score | Impressions | Eng% | Clicks | Notes
----|---------|----------------|---------------------------|-------|-------------|------|--------|-------
1   | Mon     | market_edu     | "Most buyers don't..."    | 81    | ___         | ___  | ___    |
2   | Tue     | social_proof   | "Client bought in..."     | 88    | ___         | ___  | ___    |
...

Paste into Notion or Google Sheets. After 30 days, export as JSON and pass to --past-results for an optimized next run.

How the skill works#

Uses generate_calendar.py in this directory. Local MLX first (free, unlimited), Haiku fallback.

python

# Base content mix — shifts when --past-results provided
DEFAULT_CONTENT_MIX = {
    "market_education": 0.25,   # Stats, trends, myth-busting
    "social_proof":     0.20,   # Client outcomes, testimonials
    "objection_bust":   0.20,   # Handle common objections
    "community":        0.15,   # Local spotlight, brand building
    "direct_cta":       0.20,   # Urgency-driven asks
}

# Adaptive mix: past_results → recalculate weights toward top performers
def adaptive_content_mix(past_results, base_mix=DEFAULT_CONTENT_MIX):
    if not past_results:
        return base_mix
    # Compute avg engagement rate per content_type from past results
    perf = {}
    for post in past_results["posts"]:
        ct = post["content_type"]
        perf.setdefault(ct, []).append(post.get("engagement_rate", 0))
    avg_perf = {ct: sum(v)/len(v) for ct, v in perf.items()}
    # Blend: 60% performance-weighted, 40% base (prevents over-pivoting)
    total_perf = sum(avg_perf.values()) or 1
    new_mix = {}
    for ct in base_mix:
        perf_weight = avg_perf.get(ct, sum(avg_perf.values()) / len(base_mix)) / total_perf
        new_mix[ct] = 0.6 * perf_weight + 0.4 * base_mix[ct]
    # Normalize to 1.0
    total = sum(new_mix.values())
    return {ct: v / total for ct, v in new_mix.items()}

WEEK_THEMES = [
    "Market Reality",       # Week 1 — establish authority with facts
    "Social Proof",         # Week 2 — build trust with outcomes
    "Education",            # Week 3 — deepen value with expertise
    "Urgency & Action",     # Week 4 — close with timing
]

PLATFORM_LIMITS = {
    "linkedin":  {"chars": 3000, "hashtags": 5,  "tone": "professional", "reading_level": "grade_10"},
    "x":         {"chars": 280,  "hashtags": 2,  "tone": "punchy",        "reading_level": "grade_8"},
    "facebook":  {"chars": 500,  "hashtags": 3,  "tone": "community",     "reading_level": "grade_8"},
    "instagram": {"chars": 300,  "hashtags": 10, "tone": "visual-first",  "reading_level": "grade_7"},
}

# Scoring system — runs on every generated post
POST_SCORE_WEIGHTS = {
    "hook_strength":  30,   # Pattern interrupt quality: specificity, curiosity gap, unexpected angle
    "platform_fit":   25,   # Char count, hashtag count, tone match, reading level
    "specificity":    25,   # Concrete numbers/outcomes vs. vague generalities
    "cta_clarity":    20,   # Clear single ask, low friction
}

# Auto-rewrite: posts below MIN_SCORE get regenerated with targeted fix instructions
MIN_SCORE_DEFAULT = 70
MAX_REWRITE_PASSES = 2

Core generation + scoring prompt:

python

CALENDAR_SYSTEM = """You are a social media strategist and direct-response copywriter.
Generate {days} days of {platform} content for: {niche} targeting {audience}.

Rules:
- Hook first: every post opens with a pattern interrupt — a question, stat, or bold claim
- No topic repeats within 7 days
- Match platform tone: {platform_tone}
- Content mix: {content_mix_instructions}
- Week themes: {week_themes}
- Keep under {char_limit} characters per post
- Reading level target: {reading_level}
- {tone_instruction}
- Never use: [forbidden words for niche]

Output ONLY valid JSON matching the schema provided."""

SCORER_SYSTEM = """Score this social media post 0-100 across 4 dimensions.
Return JSON: {{"hook_strength": N, "platform_fit": N, "specificity": N, "cta_clarity": N, "total": N, "fix_notes": "..."}}
Weights: hook_strength/30, platform_fit/25, specificity/25, cta_clarity/20.
fix_notes = specific instructions to fix the lowest-scoring dimension (one sentence, actionable).
Platform: {platform}. Char limit: {char_limit}."""

REWRITER_SYSTEM = """Rewrite this post. The issue is: {fix_notes}. Keep: same topic, same CTA intent, same platform.
Change: fix the specific issue identified. Output only the rewritten post JSON."""

Usage patterns#

Patrick's 2-hour content schedule method:

bash

# Step 1: Generate 30-day calendar with quality gate + tracker
python3 generate_calendar.py --niche="mortgage broker" --audience="move-up buyers" \
  --platform=linkedin --days=30 --csv --tracker --output=calendar_march.csv

# Step 2: Review scored posts in CSV — all posts already meet quality bar
# Step 3: Fill in tracker as you post, export results.json after 30 days
# Step 4: Next month — optimized calendar based on what worked
python3 generate_calendar.py --niche="mortgage broker" --audience="move-up buyers" \
  --platform=linkedin --days=30 --past-results=march_results.json --csv

Raise the quality bar:

bash

# Default min-score is 70 — set 80 for premium content
python3 generate_calendar.py --niche="SaaS startup" --audience="SMB owners" \
  --days=30 --min-score=80
# Posts below 80 get up to 2 rewrite passes — slower but higher quality

Multi-platform week:

bash

# All 4 platforms, 7 days, A/B hooks
python3 generate_calendar.py --niche="real estate" --audience="first-time buyers" \
  --platform=all --days=7 --ab-variants

Crypto trading content:

bash

python3 generate_calendar.py --niche="crypto trader + DeFi analyst" \
  --audience="retail investors" --platform=x --days=30 --tone=bold

Pipeline (JSON → scheduler automation):

bash

python3 generate_calendar.py --niche="SaaS startup" --audience="SMB owners" \
  --format=json | jq '.weeks[0].posts[] | {day, hook, cta}'

Integration with agent infrastructure#

bash

# Via Telegram
@openclaw content-calendar "30-day LinkedIn calendar for mortgage broker targeting move-up buyers"
@openclaw content-calendar "7-day X calendar, crypto niche, bold tone"

# Via Claude Code
openclaw run content-calendar --niche="real estate" --days=30 --platform=linkedin

Content pillar reference#

Pillar% of calendarWhat it doesExample hookMarket Education25%Build authority with facts"Most buyers don't know this costs them $340/month"Social Proof20%Build trust with outcomes"Client bought in [City] with $0 out of pocket. Here's how."Objection Busting20%Remove buying resistance"Waiting for rates to drop? Here's what that's actually costing you."Community15%Brand affinity, local presence"Best kept secret neighborhood in [City] right now"Direct CTA20%Drive action"Drop your zip code — I'll pull what's moving in your area today."