diff --git a/.claude/commands/auto-post-cycle.md b/.claude/commands/auto-post-cycle.md
new file mode 100644
index 0000000..97441a7
--- /dev/null
+++ b/.claude/commands/auto-post-cycle.md
@@ -0,0 +1,3 @@
+Read and follow the instructions in skills/auto-post-cycle.md
+
+The user's arguments are: $ARGUMENTS
diff --git a/CLAUDE.md b/CLAUDE.md
index a5f91c5..bdfa4a0 100644
--- a/CLAUDE.md
+++ b/CLAUDE.md
@@ -26,12 +26,16 @@ RealRate/
 │   ├── deep-dive.md              ← Build a deep dive LinkedIn document
 │   ├── content-calendar.md       ← Plan a content calendar (company page or Holger)
 │   ├── outreach-campaign.md      ← Write Instantly campaigns or LinkedIn DM sequences
-│   └── market-research.md        ← Run competitor or market research
+│   ├── market-research.md        ← Run competitor or market research
+│   └── auto-post-cycle.md        ← Generate full bizdev cycle (all 8 post types) + upload to dashboard
 │
 └── SOP/                          ← Multi-step process references
     ├── ranking-publication-protocol.md  ← Full ranking launch sequence (Day 0 → Day +10)
     ├── outreach-sop.md                  ← LinkedIn DM + Instantly detailed process
     └── research-sop.md                  ← Market and competitor research process
+│
+├── scripts/                      ← Automation scripts
+│   └── pull_seal_posts.py        ← Download seal post images + captions from Google Drive
 ```
 
 ---
@@ -57,6 +61,7 @@ RealRate/
 | Plan a content calendar | `skills/content-calendar.md` | — |
 | Write an Instantly campaign or sequence | `skills/outreach-campaign.md` | — |
 | Run competitor or market research | `skills/market-research.md` | — |
+| Generate full bizdev cycle + upload | `skills/auto-post-cycle.md` | All skills + context (auto-loaded) |
 | Full ranking launch | `SOP/ranking-publication-protocol.md` | `skills/insight-post.md` · `skills/deep-dive.md` |
 
 ---
diff --git a/scripts/pull_seal_posts.py b/scripts/pull_seal_posts.py
new file mode 100644
index 0000000..e7dbb25
--- /dev/null
+++ b/scripts/pull_seal_posts.py
@@ -0,0 +1,218 @@
+"""Download seal post images and captions from Google Drive.
+
+Usage:
+    python scripts/pull_seal_posts.py --industry us_air --output-dir output/us_air/2026
+
+Authenticates with a Google service account and downloads matching files
+from the shared seal posts Drive folder.
+
+Environment:
+    GOOGLE_SERVICE_ACCOUNT_KEY — path to service account JSON key file,
+                                  or the raw JSON content itself
+"""
+import argparse
+import io
+import json
+import os
+import re
+import sys
+
+# # Drive folder ID containing seal post assets
+DRIVE_FOLDER_ID = "1i1m4hsmOKzYrXrsPYxGrtuwMMOrQHH2T"
+
+
+def get_credentials():
+    """Load Google service account credentials from env var."""
+    key_source = os.environ.get("GOOGLE_SERVICE_ACCOUNT_KEY", "")
+    if not key_source:
+        print("ERROR: GOOGLE_SERVICE_ACCOUNT_KEY not set")
+        sys.exit(1)
+
+    from google.oauth2 import service_account
+
+    # # If it's a file path, load from file; otherwise parse as raw JSON
+    if os.path.isfile(key_source):
+        return service_account.Credentials.from_service_account_file(
+            key_source,
+            scopes=["https://www.googleapis.com/auth/drive.readonly"],
+        )
+    else:
+        info = json.loads(key_source)
+        return service_account.Credentials.from_service_account_info(
+            info,
+            scopes=["https://www.googleapis.com/auth/drive.readonly"],
+        )
+
+
+def build_drive_service(credentials):
+    """Build Google Drive API v3 service."""
+    from googleapiclient.discovery import build
+    return build("drive", "v3", credentials=credentials)
+
+
+def list_folder_contents(service, folder_id):
+    """List all files in a Drive folder (non-recursive)."""
+    files = []
+    page_token = None
+    while True:
+        response = service.files().list(
+            q="'%s' in parents and trashed = false" % folder_id,
+            fields="nextPageToken, files(id, name, mimeType)",
+            pageToken=page_token,
+            pageSize=100,
+        ).execute()
+        files.extend(response.get("files", []))
+        page_token = response.get("nextPageToken")
+        if not page_token:
+            break
+    return files
+
+
+def download_file(service, file_id):
+    """Download a file's content from Drive."""
+    from googleapiclient.http import MediaIoBaseDownload
+    request = service.files().get_media(fileId=file_id)
+    buffer = io.BytesIO()
+    downloader = MediaIoBaseDownload(buffer, request)
+    done = False
+    while not done:
+        _, done = downloader.next_chunk()
+    buffer.seek(0)
+    return buffer.read()
+
+
+def slug_from_industry(industry):
+    """Convert industry slug to human-readable folder name patterns.
+
+    us_air → ['US Air', 'us_air', 'US_Air', 'Air']
+    """
+    clean = industry.replace("us_", "")
+    return [
+        "US " + clean.title(),
+        industry,
+        "US_" + clean.title(),
+        clean.title(),
+        clean,
+    ]
+
+
+def find_subfolder(files, industry):
+    """Find a subfolder matching the industry name (Layout A)."""
+    patterns = slug_from_industry(industry)
+    for f in files:
+        if f["mimeType"] == "application/vnd.google-apps.folder":
+            for pattern in patterns:
+                if f["name"].lower() == pattern.lower():
+                    return f
+    return None
+
+
+def match_flat_files(files, industry):
+    """Match files by prefix naming convention (Layout B)."""
+    prefix_patterns = slug_from_industry(industry)
+    matched = []
+    for f in files:
+        if f["mimeType"] == "application/vnd.google-apps.folder":
+            continue
+        name_lower = f["name"].lower()
+        for prefix in prefix_patterns:
+            if name_lower.startswith(prefix.lower() + "_"):
+                matched.append(f)
+                break
+    return matched
+
+
+def extract_company_slug(filename, industry):
+    """Extract company slug from seal post filename.
+
+    seal_company_name.png → company_name
+    us_air_company_name_seal.png → company_name
+    """
+    # # Pattern: seal_<company>.png
+    m = re.match(r"seal_(.+)\.png$", filename, re.IGNORECASE)
+    if m:
+        return m.group(1).lower().replace(" ", "_")
+
+    # # Pattern: <prefix>_<company>_seal.png
+    for prefix in slug_from_industry(industry):
+        pattern = re.escape(prefix) + r"_(.+?)_seal\.png$"
+        m = re.match(pattern, filename, re.IGNORECASE)
+        if m:
+            return m.group(1).lower().replace(" ", "_")
+
+    # # Fallback: strip extension and use full name
+    base = os.path.splitext(filename)[0].lower().replace(" ", "_")
+    return base
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Download seal posts from Google Drive")
+    parser.add_argument("--industry", required=True, help="Industry slug e.g. us_air")
+    parser.add_argument("--output-dir", required=True, help="Directory to save downloaded files")
+    args = parser.parse_args()
+
+    os.makedirs(args.output_dir, exist_ok=True)
+
+    credentials = get_credentials()
+    service = build_drive_service(credentials)
+
+    print("Listing Drive folder contents...")
+    root_files = list_folder_contents(service, DRIVE_FOLDER_ID)
+
+    # # Try Layout A: look for a subfolder matching the industry
+    subfolder = find_subfolder(root_files, args.industry)
+    if subfolder:
+        print("Found subfolder: %s" % subfolder["name"])
+        seal_files = list_folder_contents(service, subfolder["id"])
+    else:
+        # # Layout B: match files by prefix in the root folder
+        print("No subfolder found, trying flat file matching...")
+        seal_files = match_flat_files(root_files, args.industry)
+
+    if not seal_files:
+        print("No seal post files found for industry: %s" % args.industry)
+        print(json.dumps({"downloaded": []}))
+        return
+
+    # # Separate images and captions
+    images = [f for f in seal_files if f["name"].lower().endswith((".png", ".jpg", ".jpeg"))]
+    captions = [f for f in seal_files if f["name"].lower().endswith(".txt")]
+
+    downloaded = []
+
+    for img_file in images:
+        company_slug = extract_company_slug(img_file["name"], args.industry)
+        print("Downloading: %s → seal_%s.png" % (img_file["name"], company_slug))
+
+        # # Download image
+        img_data = download_file(service, img_file["id"])
+        img_out = os.path.join(args.output_dir, "seal_%s.png" % company_slug)
+        with open(img_out, "wb") as f:
+            f.write(img_data)
+
+        # # Find matching caption file
+        caption_out = os.path.join(args.output_dir, "seal_%s_caption.txt" % company_slug)
+        caption_found = False
+        for cap_file in captions:
+            cap_name = cap_file["name"].lower()
+            if company_slug.replace("_", " ") in cap_name or company_slug in cap_name:
+                print("  Caption: %s" % cap_file["name"])
+                cap_data = download_file(service, cap_file["id"])
+                with open(caption_out, "wb") as f:
+                    f.write(cap_data)
+                caption_found = True
+                break
+
+        downloaded.append({
+            "company": company_slug,
+            "image": img_out,
+            "caption": caption_out if caption_found else None,
+        })
+
+    print("\nDownloaded %d seal post(s)" % len(downloaded))
+    # # Print JSON summary for the orchestrator to parse
+    print(json.dumps({"downloaded": downloaded}))
+
+
+if __name__ == "__main__":
+    main()
diff --git a/skills/auto-post-cycle.md b/skills/auto-post-cycle.md
new file mode 100644
index 0000000..dd043f1
--- /dev/null
+++ b/skills/auto-post-cycle.md
@@ -0,0 +1,213 @@
+---
+description: Generate and upload LinkedIn content for a full bizdev cycle
+arguments: industry_slug year [task1 task2 ...]
+---
+
+# Auto-Post Cycle — Orchestrator Skill
+
+Generate all (or selected) LinkedIn deliverables for an industry/year and upload them as drafts to the LinkedIn Scheduler dashboard.
+
+## Arguments
+
+- `$ARGUMENTS` is parsed as: `{industry_slug} {year} [task1 task2 ...]`
+- `industry_slug` (required) — e.g., `us_air`, `us_construction`
+- `year` (required) — marketing year (balance sheet year = year - 1)
+- Tasks (optional) — space-separated task IDs. Defaults to all. `insights` expands to `insight1 insight2 insight3 insight4`.
+
+Valid task IDs: `top10`, `article`, `seal`, `insight1`, `insight2`, `insight3`, `insight4`, `deepdive`, `insights`, `all`
+
+## Step 0 — Pull Latest Skills
+
+```bash
+cd C:\Users\User\Claude && git pull
+```
+
+This ensures all skill files, context files, and design system rules are up to date.
+
+## Step 1 — Parse Arguments and Set Up Output Directory
+
+Parse `$ARGUMENTS` into `industry_slug`, `year`, and `selected_tasks`.
+- If no tasks specified, select all 8: `top10 article seal insight1 insight2 insight3 insight4 deepdive`
+- If `insights` is specified, expand to `insight1 insight2 insight3 insight4`
+- If `all` is specified, select all 8
+
+Create output directory:
+```
+C:\Users\User\Claude\output\{industry_slug}\{year}\
+```
+
+Set variables:
+- `archive_slug` = `industry_slug` with `us_` prefix removed (e.g., `us_air` → `air`)
+- `bsy` = `year - 1` (balance sheet year)
+- `ranking_url` = `https://realrate-archive.com/{archive_slug}/{bsy}/website-ranking.json`
+- `prior_url` = `https://realrate-archive.com/{archive_slug}/{bsy - 1}/website-ranking.json`
+- `output_dir` = `C:\Users\User\Claude\output\{industry_slug}\{year}`
+
+## Step 2 — Fetch Ranking Data
+
+Fetch the ranking JSON:
+```
+GET {ranking_url}
+```
+
+Parse and extract:
+- Full company list with rank, ECR value, trend, top_rated flag, report_text
+- List of top-rated companies (where `top_rated === true`)
+- Industry name from the data
+
+If `insight2` is in selected tasks, also fetch `{prior_url}` for YoY comparison.
+
+**If fetch fails, STOP the pipeline and report the error.** Ranking data is required for all tasks.
+
+## Step 3 — Check Angle History (if generating insight3 or insight4)
+
+If `insight3` or `insight4` is in selected tasks:
+
+```bash
+curl -H "Authorization: Bearer $ADMIN_SECRET" \
+  "https://realrate-linkedin-scheduler.vercel.app/api/angles?slug={industry_slug}"
+```
+
+Parse the response to find angles used in the last 2 cycles. These angles are **excluded** from selection.
+
+## Step 4 — Generate Each Selected Task
+
+For each task, load the relevant skill file and follow its exact instructions. Do NOT hardcode captions or design rules — they come from the skill files.
+
+### Task: `top10`
+1. Read `skills/top10ranking.md`
+2. Run: `python generate_infographic.py {archive_slug}`
+3. Copy output to `{output_dir}/top10_{industry_slug}_{year}.png`
+4. Write caption following the template in `top10ranking.md` → save to `{output_dir}/top10_caption.txt`
+
+### Task: `article`
+1. Read `skills/linkedin-post.md` + `context/brand-core.md` + `context/brand-voice.md` + `context/design-system.md`
+2. Build HTML following the design system (1200×1200px canvas, Manrope font, dark background)
+3. Save HTML to `{output_dir}/article_announce.html`
+4. Export: `node posts/export.mjs` → `{output_dir}/article_announce.png`
+5. Write caption using the storytelling formula → save to `{output_dir}/article_announce_caption.txt`
+
+### Task: `seal`
+1. Run the Drive download script:
+   ```bash
+   python scripts/pull_seal_posts.py --industry {industry_slug} --output-dir {output_dir}
+   ```
+2. For each downloaded seal post:
+   - Read the existing caption from `seal_{company}_caption.txt`
+   - Web search for the company's CEO and CFO LinkedIn profiles
+   - Append C-level mentions to the caption: `cc @FirstName LastName — CEO, @FirstName LastName — CFO`
+   - Save updated caption back to `seal_{company}_caption.txt`
+3. If Drive download fails (no `GOOGLE_SERVICE_ACCOUNT_KEY`), skip seal posts and report the error.
+
+### Task: `insight1`
+1. Read `skills/insight-post.md` (Insight 1 section) + all context files
+2. Select an angle from the Insight 1 angle bank: **Reputable NTR**, **Unknown TR**, or **The Surprise**
+3. Pick a company that fits the angle. Verify its ECR data at `realrate-archive.com/{archive_slug}/{bsy}/`
+4. Build HTML using the L4 layout from `design-system.md`
+5. Save HTML to `{output_dir}/insight1_{company_slug}.html`
+6. Export PNG: `node posts/export.mjs` → `{output_dir}/insight1_{company_slug}.png`
+7. Write caption per the approved Insight 1 structure → save to `{output_dir}/insight1_caption.txt`
+
+### Task: `insight2`
+1. Read `skills/insight-post.md` (Insight 2 section) + all context files
+2. Choose: **Industry-Wide** or **Company-Level YoY Shift** based on the data (compare current vs prior year)
+3. Build up to 3 HTML slides using the L2 layout
+4. Save as `{output_dir}/insight2_slide1_{slug}.html`, `insight2_slide2_{slug}.html`, etc.
+5. Export each slide to PNG
+6. Write one caption for all slides → save to `{output_dir}/insight2_caption.txt`
+
+### Task: `insight3`
+1. Read `skills/insight-post.md` (Insight 3 section) + all context files
+2. Select a **rotating angle** from the bank: Industry Blind Spot, Market vs Balance Sheet, Industry Trend, Methodology Moment, Investor Signal, Macro Lens
+3. **Exclude** angles used in the last 2 cycles (from Step 3 angle history)
+4. Build HTML (L4 layout), export PNG
+5. Save to `{output_dir}/insight3_{angle_slug}.html` and `.png`
+6. Write caption → save to `{output_dir}/insight3_caption.txt`
+
+### Task: `insight4`
+1. Read `skills/insight-post.md` (Insight 4 section) + all context files
+2. Select a **rotating NTR angle** from the bank: Almost There, The One Thing, The Fallen, Sector Drag, Close But Not Rated, Due Diligence Gap
+3. **Exclude** angles used in the last 2 cycles (from Step 3 angle history)
+4. Build HTML, export PNG
+5. Save to `{output_dir}/insight4_{angle_slug}.html` and `.png`
+6. Write caption with mandatory ECR disclaimer → save to `{output_dir}/insight4_caption.txt`
+
+### Task: `deepdive`
+1. Read `skills/deep-dive.md` + all context files
+2. Build 8–10 slide PDF: cover → snapshot → top tier → biggest mover → surprise → warning → separators → implications → methodology → CTA
+3. Save to `{output_dir}/deepdive.pdf`
+4. Write caption → save to `{output_dir}/deepdive_caption.txt`
+
+## Step 5 — Save Angle Selections (if insight3 or insight4 was generated)
+
+If insight3 or insight4 was generated, save the selected angles to the dashboard:
+
+```bash
+curl -X POST -H "Authorization: Bearer $ADMIN_SECRET" \
+  -H "Content-Type: application/json" \
+  -d '{"industrySlug":"{industry_slug}","year":{year},"angles":{"insight3":"{selected_insight3_angle}","insight4":"{selected_insight4_angle}"}}' \
+  https://realrate-linkedin-scheduler.vercel.app/api/angles
+```
+
+Only include the angles that were actually generated (e.g., if only insight3 was requested, only send insight3).
+
+## Step 6 — Upload to Dashboard
+
+Run the upload script:
+
+```bash
+python C:\Users\User\realrate-linkedin-scheduler\upload_to_scheduler.py \
+  --industry {industry_slug} --year {year} \
+  --output-dir {output_dir} --token $ADMIN_SECRET
+```
+
+The script will find all generated files, encode them, and upload as drafts.
+
+## Step 7 — Summary
+
+Print a summary table:
+
+```
+╔══════════════════════════════════════════════╗
+║  Auto-Post Cycle Complete                    ║
+╠══════════════════════════════════════════════╣
+║  Industry: U.S. Air                          ║
+║  Year:     2026                              ║
+║  Output:   C:\Users\User\Claude\output\...   ║
+╠══════════════════════════════════════════════╣
+║  ✓ top10     — Top 10 Ranking                ║
+║  ✓ article   — Article Announcement          ║
+║  ✓ seal (3)  — Seal Posts                    ║
+║  ✓ insight1  — Company Editorial             ║
+║  ✓ insight2  — Industry YoY (2 slides)       ║
+║  ✓ insight3  — Macro Lens                    ║
+║  ✓ insight4  — The Fallen                    ║
+║  ✓ deepdive  — Deep Dive Document            ║
+╠══════════════════════════════════════════════╣
+║  Uploaded: 11 drafts                         ║
+║  Dashboard: realrate-linkedin-scheduler...   ║
+╚══════════════════════════════════════════════╝
+```
+
+Report any skipped tasks and errors.
+
+**Remind:** "Review drafts in the dashboard before approving for publish."
+
+## Error Handling
+
+- **Ranking data fetch fails:** STOP — data is required for all tasks
+- **Individual task fails:** Skip that task, continue with remaining tasks, report in summary
+- **Drive not accessible:** Skip seal posts, report "Set up GOOGLE_SERVICE_ACCOUNT_KEY"
+- **Puppeteer not installed:** Skip HTML-rendered tasks, report "Run `npm install` in `posts/`"
+- **Upload fails:** Files persist in output folder — team can retry or upload manually
+- **C-level search fails:** Keep caption without mentions, flag for manual review
+
+## Standing Rules (from brand-core.md)
+
+These rules apply to ALL generated captions and images:
+- Never use hashtags (except Top 10 which uses exactly 8)
+- Maximum 2 emoji per caption
+- Never use "we" — always "RealRate"
+- Ranking URL goes at the end of every caption
+- The archive is internal only — never link directly to it
+- Font minimum 18pt body, 46pt+ headlines (mobile readability)