Install
$ agentstack add skill-dkeken-codex-skills-alternative-creative-production ✓ scanned · ✓ verified — works with Claude Code, Cursor, and more.
Security review
✓ PassedNo issues found. Passed automated security review. · v0.1.0 How review works →
- ✓ Prompt-injection patterns
- ✓ Secret / credential exfiltration
- ✓ Dangerous shell & filesystem operations
- ✓ Untrusted network calls
- ✓ Known-malicious package signatures
What it can access
- ✓ Network access No
- ✓ Filesystem access No
- ✓ Shell / process execution No
- ✓ Environment & secrets No
- ✓ Dynamic code execution No
From automated source analysis of v0.1.0. “Used” means the capability is present in the source — more access means more to trust, not that it’s unsafe.
About
Creative Production — Orchestrator
The umbrella for the 8 creative-* skills. Your job: read the request, locate it in the creative process, and route to the right sub-skill. Don't do the work here — dispatch. Think of yourself as a creative director assigning the task, not executing it.
When to use
- The user gives a broad creative goal ("make campaign assets", "I need visuals for the launch") and the specific step is ambiguous.
- A creative project spans multiple stages and needs sequencing.
- The user doesn't know which sub-skill they need.
When NOT to use
- The user explicitly named a step ("write me 25 ad concepts") → go straight to that sub-skill.
- The task is product-design/prototyping, not visual asset creation →
product-designorchestrator.
The pipeline
DIRECTION ASSETS FINISH
creative-explore ─────┐
creative-moodboard ───┼──► creative-scene ───────┐
creative-positioning ─┘ creative-shot ├──► creative-polish
creative-ads-explorer │
creative-offer ──────────┘
Direction is locked first, assets are generated against it, winners are polished last.
Routing table
| User intent / phrasing | Route to | |---|---| | "what style / which look", no direction locked | creative-explore | | "align the team", "reference board", "territories" | creative-moodboard | | "make visuals match our strategy/positioning" | creative-positioning | | "sell this offer/deal/bundle", "promo creative" | creative-offer | | "ad concepts", "variations to test", "25 ads" | creative-ads-explorer | | "product in use", "lifestyle shot", "show it being used" | creative-scene | | "different angles", "PDP/ecommerce set", "shot list" | creative-shot | | "finalize", "clean up", "production-ready", "fix artifacts" | creative-polish |
How to orchestrate
- Check for a locked direction. If none exists, start with
creative-explore(single look) orcreative-moodboard(team alignment) orcreative-positioning(strategy-led). Never generate final assets against an unlocked direction. - Route asset generation to
scene/shot/ads-explorer/offerbased on the deliverable. - Always finish deliverables through
creative-polish. - Carry the locked direction + prompt stub forward so every sub-skill stays consistent — this is what makes a set look like one campaign.
Worked example
Request: "We're launching a coffee subscription. I need everything — a look, a hero, some ads, and a clean set for the site."
plan:
1. no look locked → creative-explore → user picks "Warm Ritual", stub saved
2. hero lifestyle → creative-scene (reuse stub) → "morning pour" hero
3. ad test batch → creative-ads-explorer (reuse stub) → 12 hypotheses
4. site product set→ creative-shot (reuse stub) → 6-angle PDP set
5. finish winners → creative-polish → final hero + top 3 ads + PDP set
dispatch order respects: direction → assets → polish
Quality bar
- A direction is locked before any final asset is produced.
- Every dispatched sub-skill receives the same locked direction/stub (coherent set).
- The user sees a clear plan (which skills, in what order) before execution on multi-stage requests.
Common pitfalls
- Doing the work here — this skill decides WHICH skill; the sub-skill executes. Don't generate images in the orchestrator.
- Skipping direction-lock to save time — costs far more in inconsistent output later.
- Losing the stub between stages — leads to a set that looks like 5 different brands.
Tooling
Image generation happens inside the sub-skills via a text-to-image tool. If none is available, the sub-skills return prompts to run manually; the orchestrator still sequences the work.
Source & license
This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.
- Author: DKeken
- Source: DKeken/codex-skills-alternative
- License: MIT
Install and usage instructions live in the source repository linked above.
Reviews
No reviews yet — be the first.
Write a review
Versions
- v0.1.0 Imported from the upstream source.