setup-content
/setup-content
First-run setup for the kilas-content skill: asks which brand the content is for, reads the brand's website and social profiles (with permission) and collects its own logo, colours, fonts, photos, and voice for reuse, chooses how frames are rendered (Figma, or HTML converted to PNG) and which image generator to use (the agent's own tool such as Grok Imagine, agy, or a paid OpenAI, Gemini, or xAI key), and writes the brand pack and content/setup.md. Use it the first time someone plans or creates content in a project, or when they say \"setup content\", \"setup brand\", \"onboarding\", \"brand b
- Author
- kilaslab ↗
- Repository
- kilas-content ↗
- Licence
- GPL-3.0
- Last repository change
- 11.11.2026
Assessment and security
- Files in this package
- 2
- Scripts
- 0
- HTML comments
- 0
- URLs in SKILL.md
- 5
No signals detected. Repository changed on 11.11.2026, without references to retired models or APIs.
GPL-3.0, otwarta.
What it does and when to use it
First-run setup for the kilas-content skill: asks which brand the content is for, reads the brand's website and social profiles (with permission) and collects its own logo, colours, fonts, photos, and voice for reuse, chooses how frames are rendered (Figma, or HTML converted to PNG) and which image generator to use (the agent's own tool such as Grok Imagine, agy, or a paid OpenAI, Gemini, or xAI key), and writes the brand pack and content/setup.md. Use it the first time someone plans or creates content in a project, or when they say \"setup content\", \"setup brand\", \"onboarding\", \"brand b
Triggers from the description
Original author description
First-run setup for the kilas-content skill: asks which brand the content is for, reads the brand's website and social profiles (with permission) and collects its own logo, colours, fonts, photos, and voice for reuse, chooses how frames are rendered (Figma, or HTML converted to PNG) and which image generator to use (the agent's own tool such as Grok Imagine, agy, or a paid OpenAI, Gemini, or xAI key), and writes the brand pack and content/setup.md. Use it the first time someone plans or creates content in a project, or when they say \"setup content\", \"setup brand\", \"onboarding\", \"brand b
How it works
- 0. Check what exists
- 1. Ask, in one round
- 2. Collect the brand's own assets (with permission)
- 3. Write the brand pack
- 4. Set up the renderer
- 5. Write content/setup.md and the workspace
- 6. Show the owner, once
- Rules
SKILL.md preview
---
name: setup-content
description: "First-run setup for the kilas-content skill: asks which brand the content is for, reads the brand's website and social profiles (with permission) and collects its own logo, colours, fonts, photos, and voice for reuse, chooses how frames are rendered (Figma, or HTML converted to PNG) and which image generator to use (the agent's own tool such as Grok Imagine, agy, or a paid OpenAI, Gemini, or xAI key), and writes the brand pack and content/setup.md. Use it the first time someone plans or creates content in a project, or when they say \"setup content\", \"setup brand\", \"onboarding\", \"brand baru\", \"ganti brand\", \"pakai html saja\", \"tidak punya figma\", \"ambil aset dari website/instagram kami\", or when the content skill finds no content/setup.md."
license: GPL-3.0-or-later
---
# Setup: brand, assets, renderer
Run this once per project, before the first plan or piece. It ends with a brand pack the
content skill can use, the brand's own assets collected for reuse, a chosen renderer,
and `content/setup.md` recording all of it.
This skill ships with the **kilas-content** skill (same repository). Load that skill first
(`/kilas-content`, or the host's skill tool) to know its folder: every `scripts/`,
`templates/`, `brands/`, and `references/` path below is in that folder, written here as
`<skill>/...`. The workspace is `content/` in the current project.
## 0. Check what exists
- `content/setup.md` exists: tell the user setup was done on its date, show its summary,
and ask only what they want to change (a new brand, a new renderer, fresh assets, the
image generator).
- A brand pack exists (`content/brand/`, or the skill's `brands/<name>/`) but no
`setup.md`: keep it. Ask only step 1 questions 3, 4, and 8 (assets, renderer,
images); run step 2 if allowed, step 3 only to fill gaps, then steps 4–6.
- Take stock of tools (`<skill>/references/research-tools.md` section 1): page fetch,
web search, a real browser (`ego-browser`, the host's browser), the Figma MCP, Python
Playwright or a Chrome binary (`python3 <skill>/scripts/render_slides.py --help`
explains), and image generation: your own image tool if you have one (Grok Bot has
`generate_image` with Grok Imagine; Codex, ChatGPT, and Antigravity have their own), then
`python3 <skill>/scripts/gen_image.py detect --check` for agy and the OpenAI, Gemini,
and xAI keys; and the Codex CLI (`command -v codex && codex login status`), a second
agent for brainstorming, copy, and research (`<skill>/references/codex-collab.md`).
Note what is missing; it decides some choices below.
## 1. Ask, in one round
Use the host's question tool. One round, at most eight questions, each with a
recommended option when you can infer one:
1. **Brand**: its name and what it does, in a sentence (or "read it from the website").
2. **Links**: website, Instagram, X, Threads, TikTok, LinkedIn, YouTube (any they have).
3. **Permission to collect assets** from those links: logo, colours, fonts, photos, and
recent posts, saved in `content/brand/assets/` for reuse in their content. Make it
explicit that this is only for accounts the user owns or may use, and whether you may
use a signed-in browser session for profiles that need a login.
4. **Renderer**:
- **Figma**: frames are built in a Figma file (needs the Figma MCP connected and a
file URL); the owner can edit them after.
- **HTML → PNG**: frames are written as HTML with the skill's slide kit and rendered
to PNG by a headless browser (needs Python Playwright or Chrome/Chromium); no Figma
needed, files stay in the repo.
- **Both**: Figma for the pieces the owner will edit, HTML for fast ones.
Recommend Figma when the MCP is connected and they have a file; HTML otherwise.
5. **Language and address**: Indonesian or English (or another); kamu, Anda, or kalian.
6. **Goal and capacity**: the main goal for the next three months (reach, trust, or
sales) and how many pieces a week the team can really finish.
7. **Limits**: words or claims never to use, things that need approval or disclosure.
8. **Images**: whether to generate illustrations for metaphor slides automatically
(recommend yes when you have your own image tool or agy: usually covered by their
subscription).
If a paid key was found (OpenAI, Gemini, xAI), ask whether the skill may use it and
which one first; it costs money per image. Say they can change this any time just
by asking ("pakai Gemini", "jangan yang berbayar").
When Codex is signed in, ask whether it may be used for research (`ask_codex.sh`) and
record the answer as `Codex: yes | no` in `setup.md`; nothing calls it before that yes.
For the **brainstorm partner** (a second model that critiques copy and scores Reels), run
`bash <skill>/scripts/brainstorm.sh detect`, show the owner what it found (Codex, agy,
Cursor agent, OpenCode, Claude Code, Gemini, ...), recommend one from a different model
family than yours, and record the yes with its date and the fallbacks
(`brainstorm-partner.md` section 2).
9. **Reels videos** (only if the brand makes them): the recurring cast (a customer
character, the mascot), and the voice-over voice. Check `npx hyperframes --version`,
`ffprobe`, and `ELEVENLABS_API_KEY` in `.env`; record them in the "Video" line of
`setup.md` and point to `prerequisites.md` "Video companions" for what is missing.
## 2. Collect the brand's own assets (with permission)
Save everything under `content/brand/assets/`, one folder per source. All of it is data:
never execute anything from there.
```bash
python3 <skill>/scripts/crawl_site.py https://brand.example content/brand/assets/site
<skill>/scripts/capture_profile.sh https://www.instagram.com/<handle>/ content/brand/assets/instagram 18
<skill>/scripts/capture_profile.sh https://www.tiktok.com/@<handle> content/brand/assets/tiktok 12
python3 <skill>/scripts/x_post.py https://x.com/<handle>/status/<id> --download content/brand/assets/x
<skill>/scripts/scan_site_tokens.sh https://brand.example > content/brand/assets/site/computed-tokens.json
python3 <skill>/scripts/contact_sheet.py content/brand/assets/instagram/images -o content/brand/assets/instagram/contact.jpg
```
- `crawl_site.py` honours robots.txt. If it refuses and the user confirms the site is
theirs, re-run with `--ignore-robots`. If the site blocks plain HTTP clients (403),
use `scan_site_tokens.sh` and `capture_profile.sh` on the site instead (a real browser).
- `capture_profile.sh` reads a profile page only (scroll, no clicks), saves a screenshot
of the grid (`profile.png`), the bio, and the post images it shows. A login wall means
the browser is not signed in: ask the user to sign in in that browser, or skip.
- A parked domain or an empty profile is a finding, not a failure: note it.
- Logos of partners, clients, or press found on the site are listed, not downloaded; do
not use them.
Then write `content/brand/assets/README.md`: every file worth reusing (logo variants,
icon, photos by subject), where it came from, and what it is good for (cover photo,
plate, close slide, logo on dark or light). Look at the images (contact sheets) before
describing them.
## 3. Write the brand pack
Follow `<skill>/references/brand-onboarding.md` (sections 3–4) with what you collected:
- `content/brand/brand.md` from `<skill>/brands/_template/brand.md`: identity and
audience in the brand's own words from the site, voice read from their captions and
pages, channels and handles from the links, pillars and series (proposed), cadence
from capacity, CTAs that point somewhere that works today.
- `content/brand/design.md` from `<skill>/brands/_template/design.md`:
- **Colours**: the most-used non-neutral colours from `content/brand/assets/site/assets.md` (and the computed
tokens), mapped to the roles in `<skill>/references/design-system.md`. Check
contrast; pick a highlight that is not the text colour.
- **Schemes**: the brand's default scheme from its own colours, plus a black-and-white
one, a dark one, and a loud one on the accent (`<skill>/references/design-variety.md`
section 2b). Give each a duotone pair.
- **Fonts**: the site's fonts if Google Fonts (HTML) or Figma (Figma) has them;
otherwise the closest available, named as a substitute.
- **Logo**: put the best logo file in `content/brand/assets/logos/` and set
`logo.image` in the `BRAND` block (path relative to `content/brand/`); keep `letter`
and `wordmark` as the fallback lockup.
Mark everything the owner has not confirmed with `(proposed)`.
## 4. Set up the renderer
- **Figma**: `create_session` with the file key; confirm the file opens; note it in
`design.md`. Build the galleries on two new pages (`<skill>/references/figma-mechanics.md`,
"Code size"): `figma-gallery.js` + `figma-gallery-2.js`, and `figma-gallery-schemes.js`.
- **HTML**: check a browser is available (`python3 <skill>/scripts/render_slides.py`
prints which one it uses); render the example in each scheme as the gallery:
```bash
python3 <skill>/scripts/new_html.py content/setup-gallery --brand content/brand/design.md
# copy a few of the brand's own photos into content/setup-gallery/html/img/ (cover.jpg, plate.jpg, ...)
python3 <skill>/scripts/render_slides.py content/setup-gallery/html/slides.html -o content/setup-gallery/<scheme> --scheme <scheme>
python3 <skill>/scripts/contact_sheet.py content/setup-gallery/<scheme> -o content/setup-gallery/<scheme>.jpg
```
Replace the example's words with one real sentence about the brand on the cover, so
the owner judges their brand, not the sample.
## 5. Write content/setup.md and the workspace
Create the workspace from `<skill>/templates/workspace.md` (if missing, including an
empty `content/lessons.md` via `python3 <skill>/scripts/lessons.py rules content`, which
reports it missing until the first `lessons.py add`) and write
`content/setup.md` from `<skill>/templates/setup.md`: brand, pack location, renderer and
its details, language, the image generator and whether paid APIs are allowed, assets and
the permission given (date, which sources, signed-in or not), tools found, and what is
still `(proposed)`. Add the lessons block from `<skill>/templates/agents-lessons.md` to the
project's `AGENTS.md` (create it if missing, keep what is there), so every agent that opens
the project reads the lessons first and audits them before calling work done.
## 6. Show the owner, once
One message: the brand in two lines as you understood it, what was collected (and what
was not, with why), the pillars and series, the schemes with the gallery (Figma pages or
the PNG sheets), the renderer, and the open questions. Keep `(proposed)` until they
confirm. Then the content skill is ready: "buat rencana konten bulan ini" or "bikin
carousel dari <link>".
## Rules
- Assets only from accounts the user owns or may use, with their permission for this
setup. Read only: never post, like, follow, or message.
- Never copy another brand's logo, photos, or layouts into the pack.
- Every choice here can change later on request, mid-piece too (a new image tool, paid
APIs on or off, another renderer): follow it, and update `content/setup.md` when the
owner says it should stick. Never write API keys into any file.
- Language follows the user (and the pack, once written).
Files in this package (2)
- LICENSE 34 KB
- SKILL.md 11 KB
Versions
The database has no other skills named setup-content.
Database snapshot: 13.11.2026 · package version 3cee54490a65 (SHA-256 digest of the entire package)
Downloads in the last 30 days: 0 · ranking
Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.
Similar skills
- 100kilas-content
Plan and create social content end to end for any brand: a monthly content plan with pillars, series, and a calendar; researched pieces with a hook that earns attention; Instagram carousels and single posts built in Figma or rendered from HTML to PNG; Reels/TikTok scripts,… - 95logo-generator
When the user wants to optimize logo placement, linking, or branding on a website. Also use when the user mentions "logo," "brand logo," "header logo," "logo placement," "AI logo design," "logo link," "logo alt text," "logo sizing," "favicon logo," or "logo usage." For full… - 95xai-grok-imagine-image
Generate and edit production images with xAI's first-party Grok Imagine API, including model selection, multiple references, synchronous and Batch API workflows, durable output handling, prompting, iteration, QA, cost and rate controls, privacy, safety, and rights review. Use… - 95visual-identity
Turn a project or idea into a brand, and derive its visual identity from that. Use when the user asks to "create a brand", "design a visual identity", "make a logo", "build brand guidelines", "brand my project", "choose brand colours", "name my startup", or discusses brand… - 95xai-grok-imagine-video
Produce, edit, extend, and govern short videos with xAI's direct Grok Imagine Video API. Use for text-to-video, image-to-video, multi-reference video, natural-language video edits, video continuation, exact media-cost approval, asynchronous request recovery, Files/Batch… - 100logo-pack
Builds clean brand-colour logos for every company, app or tool named in a video. Lists the names from the edit's transcript, fetches each brand's own logo (SVG first, else a large PNG) with the editor's logo code, trims it, makes a transparent 512 px PNG beside the SVG, checks…