Work Product Command Center

CEO Gap Report

The short boss read, v1.3 resynthesis, grounded in 2,011 mapped products and a 39-card scorecard. Built 2026-07-02.

CEO Gap Report — Work Product Command Center

Version: v1.1 Built: 2026-06-25 Palette: The Darjeeling Limited (Wes Anderson)

This is the short boss read. Plain words, no fluff.

The numbers

What is currently working

The make-things muscle is strong. 2,011 work products are mapped from your real drives, 369 are shipped live to the website, and a deep skill library plus a working memory system steer the daily output. The quiet star is n8n. The launchd service com.drwu.n8n is running this minute with 5 robots active. Of the 39 plans, methods, and tools scored against on-disk evidence, the 4 green cards are the pieces that truly run: Obsidian Dataview live inside 22 dashboards, the n8n daemon, its 6-spec workflow set, and the Claude Code 3-file memory system. When Brandon builds, things get BUILT.

What keeps getting repeated

Planning repeats and acting stalls. 9 of the 39 scored cards are written Claude Code adoption plans, and all 9 scored red for one reason: the plan sits on disk with zero Phase 1 action. No connector added, no starter file created, never queued. The plans board says the same, 12 plans written and none enqueued. Links get saved and transcripts get captured, then a fresh batch arrives before anything ships from the last one. The missing move is identical across the reds, do only Phase 1, one file or one click.

What keeps getting lost

Finished work dies at the ship step. The inventory shows 369 shipped against 973 red, and the reds are un-started ideas and un-flipped switches. The morning-briefing robot 04-morning-briefing.json is written and was never imported, the one file between you and a finished six-robot set. Work gets lost in the last five minutes before shipping. This Command Center proved the rule on itself: it was rebuilt on June 25 and left un-synthesized and un-shipped until this pass finished it.

What tools are underused

Six tools scored red for a zero on-disk footprint: Zotero, DEVONthink, Raycast, Airtable, Figma, and Descript are named in the app-stack doc and appear nowhere on the machine. Each needs a keep-or-drop decision, not a build. Adopt one for a week, Raycast with a single quicklink to the Command Center is the cheapest test, or strike it from the stack doc so the list stops claiming a tool you never touch.

Where Obsidian would help first

The vault is built right, with numbered folders, hub pages, and Dataview live in 22 dashboards, one of your 4 greens. The first win is a daily habit. Open Workflow Command Center once each morning and start chats with /chat-start so session notes land in the vault by themselves.

Where n8n (Innate In) would help first

Import 04-morning-briefing.json and flip it on. That one file finishes the original six-robot set in about five minutes, and it is the exact step named across three separate cards. The service is already alive with a confirmed PID, so this is a switch to flip.

Where Claude Code should automate link gathering

Five red cards are saved links and transcripts with zero derived artifacts: the Cowork courses, the AI-agent courses, the Karpathy knowledge base, the n8n MCP server, and Ableton automation. Point every saved X or YouTube link at the research drop-zone robot that already runs, then let the weekly dashboard refresh score new links into the Link Plans page so nothing rots in a tab bar again.

Where browser bookmarks should be fixed

The six dashboard tabs were opened in Chrome and Firefox and ADD_TO_BOOKMARKS.txt spells out the steps. What remains is pressing Cmd+D on the Start Here tab in each browser. Ten seconds, and the Command Center is one click away forever.

Do this in the next seven days

  1. Day 1: Trim MEMORY.md back under its 24.4KB limit. It reads 26.7KB now, so the oldest entries drop out of every session in silence. One editing pass restores the memory system that steers all the work.
  2. Day 2: Import 04-morning-briefing.json into n8n and switch it on. Finishes the six-robot set in five minutes.
  3. Day 3: Pick one red Claude Code plan and build only its Phase 1 file. The Todoist connector is fastest: Settings, then Connectors, Add, Todoist, log in.
  4. Day 4: Publish one already-written, unshipped piece to one channel. Choose the piece that is closest to done and send it.
  5. Day 5: Adopt-or-strike one dead tool. Install Raycast with a single quicklink to the Command Center, or delete it from the stack doc.
  6. Day 6: Press Cmd+D on the Start Here tab in Chrome and Firefox so the Command Center stays one click away.
  7. Day 7: Attach a weekly launchd plist to scan_inventory.py and build_dashboards.py, and always run write_ceo.py before build_dashboards.py so this report never returns blank again.

Built by build_dashboards.py v1.1. Re-run after scan_inventory.py to refresh.