Menu
Start or stop the blog pipeline Kanban dashboard (Flask app at dashboard/app.py, served on http://localhost:5001).
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
Manage the blog.postman.com editorial calendar — list, reschedule, view monthly counts, or get a YTD summary. Enforces Tue/Thu-first scheduling with no conflicts or US public holidays.
Show the number of blog posts published on blog.postman.com between two dates. Prompts for a start and end date, then queries WordPress and displays the count with a post listing, broken down by month and author.
Copy edit blog post output for grammar, syntax errors, repetitive sentence structures, and SEO optimization. Suggests an SEO-friendly title and 150-character meta description. Runs automatically as a hook after blog-write, or standalone via /blog-copyeditor [filename].
Convert a Google Doc to a blog-ready markdown file with images. Takes a Google Docs URL, exports the content as markdown, downloads embedded images, and saves everything to blog-output/.
Clean up the blog pipeline Kanban dashboard — removes cards that are stuck before the staging column (ideas, creating substages). Leaves cards in staging, review, and scheduled untouched.
Generate a Postman-branded blog header image (2560×1355 PNG, no text) using the Gemini image generation API (nanobanana). Analyzes reference images and follows the Postman design system for consistent brand visuals.
| name | blog-dashboard |
| description | Start or stop the blog pipeline Kanban dashboard (Flask app at dashboard/app.py, served on http://localhost:5001). |
| argument-hint | start|stop |
| allowed-tools | ["Bash"] |
Start or stop the local Flask dashboard that visualizes the blog pipeline Kanban board. The dashboard itself never runs skills directly — it just watches blog-output/ and dashboard/state.json and spawns agents via queue.json. See dashboard/README.md for the full architecture.
Resolve <plugin-root> the same way other dashboard-touching skills do: the directory containing dashboard/app.py, either the current plugin root or CLAUDE_PLUGIN_ROOT if set.
The action is the first argument (start or stop). If missing or not one of those two, ask the user which one they meant — do not guess.
Check if it's already running:
lsof -i :5001 -sTCP:LISTEN -t
If this returns a PID, tell the user the dashboard is already running at http://localhost:5001 and stop — do not start a second instance.
Ensure the virtualenv exists. If <plugin-root>/dashboard/.venv is missing, create it and install deps:
cd <plugin-root>/dashboard && python3 -m venv .venv && .venv/bin/pip install -r requirements.txt
Launch the app in the background, logging output and capturing its PID so stop can find it later. Put the cd on its own line — chaining it as cd dir && nohup ... & backgrounds the whole cd && nohup pair in a subshell, so the following echo $! line runs in the wrong directory and writes the pidfile to the wrong place:
cd <plugin-root>/dashboard
nohup .venv/bin/python app.py > .dashboard.log 2>&1 &
echo $! > .dashboard.pid
Run this with the Bash tool's run_in_background unset (it's already detached via nohup ... &) — just run it and move on, don't block on it.
Give the server a moment to bind, then confirm:
sleep 1 && lsof -i :5001 -sTCP:LISTEN -t
If a PID comes back, tell the user the dashboard is running at http://localhost:5001. If not, read the last ~20 lines of dashboard/.dashboard.log and show the user what went wrong (commonly: missing venv deps, or missing WP_APP_PASSWORD/GEMINI_API_KEY env vars — see dashboard/README.md prerequisites).
Find the listening process:
lsof -i :5001 -sTCP:LISTEN -t
If nothing is returned, tell the user the dashboard isn't running and stop.
Kill it:
kill $(lsof -i :5001 -sTCP:LISTEN -t)
Clean up the pidfile if present:
rm -f <plugin-root>/dashboard/.dashboard.pid
Confirm to the user that the dashboard has been stopped.
lsof, not just the pidfile — the pidfile can go stale if the process was killed out-of-band or the machine restarted.dashboard/state.json, queue.json, or blog-output/ from this skill — that's /blog-dashboard-cleanup's job, and normal pipeline operation otherwise.