- Home
- Skills
- Social Media
- Showrunner: Plan, Write and Check Creator Content
Works with the AI Agents you already use
Showrunner: Plan, Write and Check Creator Content
Asks what you make, for whom and where, then plans to your real weekly hours and drafts hooks, scripts and captions for YouTube, Instagram, TikTok and LinkedIn. Never posts.
Free
Showrunner: Plan, Write and Check Creator Content
Example session with this skill installed
I want to grow my channel. Short videos for college students who want to cook cheap food in a hostel. I'm starting from zero, I have 5 hours a week including editing, I'm thinking Instagram and YouTube, and I don't want to show my face.
- Read your context and instructions
- Compiled the showrunner
Summary
The intake picked YouTube as the hub and Instagram as the spoke, parked TikTok for later because 5 hours a week fits two platforms, and planned faceless formats. This is the real output of the planning step, unedited.
INTAKE RESULT
Making: short_video | For: students | Goal: Grow an audience or personal brand | Stage: Starting from zero | 5 h/week
RANKING (fit 0 to 100)
91.3 TikTok later a strong fit, but 5 hours a week fits 2 platforms; one of the strongest fits for your goal (grow an audience or personal brand)
84.0 YouTube HUB your audience is concentrated here; heavy for 5 hours a week
79.0 Instagram spoke you asked for it
53.8 LinkedIn skip average fit across the board
47.5 Facebook skip your audience is thin here
46.2 Threads skip not a natural home for that format
40.0 X skip not a natural home for that format; your audience is thin here
Hub: youtube Spokes: instagram Skip for now: tiktok, linkedin, facebook, threads, x
Weekly load: 231 min planned of a 255 min creation budget (+45 min for replies and DMs)
youtube short_video 3 per week (~56 min each, made fresh)
instagram short_video 3 per week (~21 min each, adapted from the hub piece)
FIRST 7 DAYS
Day 1 Save your profile and capture how you talk, so every draft sounds like you. [sr-voice]
Day 1 Audit and fix the YouTube profile: name, one-line promise, link, pinned post. [sr-profile-audit]
Day 2 Build a bank of 30 ideas from your three pillars, scored. [sr-ideas]
Day 3 Collect 10 to 20 posts that beat their own account's median in your niche, then rank them. [sr-outliers]
Day 4 Write and score three hooks per idea for the best three ideas. [sr-hook]
Day 5 Script, caption and clean the first batch of pieces. Run the policy check. [sr-script]
Day 6 Lay out the first four weeks inside your hours budget. [sr-calendar]
Day 7 Post the first piece. Open an experiment log so what you learn sticks. [sr-growth-loop]
- You prefer not to be on camera. Plan faceless formats: voiceover with screen recording or b-roll, text on screen, or carousels.
- facebook, instagram, tiktok, youtube are all vertical-video platforms. Make once, then post the native version on each: new first frame, platform captions, no watermarks from other apps.
Starting hypothesis from priors in data/fit.json, not a prediction. Replace it with what your own numbers show after 6 to 8 posts.
Next steps
- Say yes to save the profile and start the zero-to-first-1000 playbook. Every step waits for your approval.
- You film, upload and post. Showrunner never posts for you.
Connects securely to your tools. The creator never sees your data.
About this skill
Most content advice an AI gives you is confident guessing: post every day everywhere, here is the best time, the algorithm loves this. Showrunner does the opposite. It asks first, sizes the plan to the hours you really have, and keeps facts apart from guesses.
1. Intake before advice. The conductor asks seven questions, at most two at a time and in your language (tested in Hinglish): what you make, who it is for, your goal, where you post, how big you are, your real weekly hours, and whether you will be on camera. Then it picks one hub platform, one or two spokes that reuse the hub piece, and what to skip for now, with the minutes it all adds up to.
2. Gated playbooks. Seven playbooks (zero to first 1,000, weekly sprint, one long video into a week of posts, YouTube launch, profile makeover, monthly review, campaign) chain the skills. Every step has a gate: it shows the work and waits for your yes.
3. The work itself. Hooks, scripts, titles and thumbnails, captions, carousels, text posts and threads, cut lists with ready-to-run ffmpeg commands, repurposing, a calendar with an .ics export, replies, and a humanizer for AI-sounding drafts. It uses details you give it, and asks for the number, the name and the mistake instead of inventing a story.
4. Honest numbers. 43 platform facts, each with a status (official, reported, inferred or study), a source and a date. Hook, title and caption scores are labeled as heuristics. A calibrate tool tests them against your own posts and tells you when a score is no better than a coin flip.
Safe by design
- It never posts, schedules or sends anything. Asked to post, it hands you the upload steps.
- No network calls and no dependencies. The engine is Python standard library.
- It writes only to
~/.showrunner(your profile, playbook runs and experiment logs), and only when you save something.
Also free on GitHub. The same skills are MIT licensed at github.com/Rajbharti06/showrunner-agent-skills, with a Claude Code plugin, zips for ChatGPT and the Claude app, and an MCP server.
How it was tested: 85 tests in CI on Linux, Windows and macOS, and a live Claude Code session on Windows with a faceless creator chatting in Hinglish. That session found six bugs, all fixed in 0.1.1.
Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Social Media.
Scores are text heuristics, not validated against views. Use calibrate on your own posts.
The hook scorer's dealbreaker patterns are English only: a Hindi greeting is not caught.
15 of the 43 platform facts do not have a verified source link yet.
No live platform data: it does not pull your analytics or trends. You bring the exports.
It never posts. Publishing is yours.
Installs not tested on macOS, and not tested inside ChatGPT, Codex, Cursor or Gemini CLI.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before buying a skill.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Free to use
No account required to browse
Trust & safety
Security scan
Security scan passed 13 days ago
- Free with an Agensi account
Nothing to install. It runs in your AI agent.
- Optional
Python 3.9+FFmpeg
Python runs the scoring, planning and clip tools. FFmpeg only runs the clip commands it writes.
Permissions
File Scopes
Reads files you give it (transcripts, CSV exports, drafts). Writes only to ~/.showrunner when you save a profile, start a playbook or log an experiment. No outbound network access: the optional MCP server listens on 127.0.0.1 only. Environment variables are optional settings such as SHOWRUNNER_HOME, never secrets. Never posts.