7.2 KiB
Creator Mode
1. Purpose
Creator Mode is used to create new CCPE artifacts.
It supports creating:
CCPE-Lite Prompt Cards
CCPE-Agent Specs
CCPE-Skill Specs
CCPE-Runtime Specs
Model Cards
Model Index entries
Hybrid artifacts
Creator Mode should not assume that every new request requires a full Agent or Runtime.
Creator Mode should also not assume that every requested prompt belongs in CCPE. Use the CCPE value gate first: accept work only when it forges expert agent capability, cognitive model structure, reusable method design, durable invocation contracts, high-risk cognitive boundaries, or cross-project reuse. Return project-local prompt glue, one-off formatting, immature model trials, sample runs, and project invocation decisions to the owning project repository.
The goal is to create the lightest artifact that preserves function, depth, maintainability, and safety.
2. When to Use Creator Mode
Use Creator Mode when the user asks to:
Create a new Agent
Create a new Skill
Create a new Prompt Card
Create a new Runtime
Create a new Model Card
Create a new Model Index entry
Design a new committee
Design a new review workflow
Turn a model into an agent
Turn a model into a skill
Turn a process into a runtime
3. Creator Mode Workflow
Follow this sequence:
1. Intake
2. CCPE Value Gate
3. Scenario Probe
4. Intended Use Analysis
5. Target Platform Analysis
6. Layer Selection
7. Classification
8. Operating Mode Assessment
9. Depth vs Automation Assessment
10. Model / Skill / Runtime Check
11. Creation Brief
12. Proposed Files
13. Artifact Draft
14. Validation
15. Final Notes
3.1 Scenario Probe
Before selecting artifact layers, determine:
Is this for Web / GPT / Gemini / Claude direct use?
Is this for Codex automatic invocation?
Is this a durable workflow role?
Is this a reusable method?
Is this a multi-agent workflow?
Will the human manually coordinate multiple agents?
Does the user expect automation, or only structured expert thinking?
If the scenario is single-agent expert use, default to CCPE-Lite. Add Skill only when Codex should invoke the method automatically. Add Agent Spec or Runtime only when collaboration, handoff, state, or automation requires it.
4. Intended Use Analysis
Determine what the new artifact is for.
Ask or infer:
What work should this artifact perform?
Who will use it?
What input will it receive?
What output should it produce?
Will it be used once or repeatedly?
Will it be used alone or inside a workflow?
Will it be used by a human, an agent, or another system?
5. Target Platform Analysis
Determine where the artifact will run.
Possible targets:
Custom GPT
Gemini Gem
Claude Project
Claude Code
Codex
OpenClaw
General Markdown spec
Platform-neutral protocol
Platform affects structure.
Examples:
Custom GPT / Gem:
Prefer portable CCPE-Lite unless complex.
Codex:
Use AGENTS.md, Skill, templates, and repository structure.
Claude Code:
Use CLAUDE.md, subagents, Skills, and project files.
OpenClaw:
Consider Agent / SubAgent / Skill structure.
Platform-neutral:
Create spec first, implementation later.
6. Classification
Classify the artifact as:
CCPE-Lite
CCPE-Agent
CCPE-Skill
CCPE-Runtime
Model Card
Model Index
Hybrid
If Hybrid, identify components.
Example:
New Cognitive Imaging Expert
= Agent role
+ Cognitive Imaging Model
+ Cognitive Imaging Skill
+ portable Lite version
+ optional committee Runtime node
7. Operating Mode
Determine whether the artifact operates as:
Expert Mode
Workshop Mode
Automation Mode
Hybrid Mode
Use Expert Mode for single-role expert interaction.
Use Workshop Mode for predefined multi-agent collaboration.
Use Automation Mode for stable, verifiable procedures.
Use Hybrid Mode when deep work combines with automated support.
8. Depth vs Automation
Determine:
Depth-Oriented
Automation-Oriented
Hybrid
Depth-Oriented artifacts need human judgment and should not be forced into full automation.
Automation-Oriented artifacts need authority, validation, and recovery rules.
Hybrid artifacts need both human decision gates and automation boundaries.
9. Cognitive Model Check
Determine whether the artifact depends on a cognitive model.
Ask:
Is there a named model?
Is the model user-authored?
Does the model have assumptions, mechanism, scope, and failure modes?
Is the model reusable outside this artifact?
Should it become a Model Card?
Should its procedure become a Skill?
If the model is important and reusable, create or propose:
Model Card
Skill
Agent that references the model
Portable Lite version if needed
10. Skill Check
Determine whether the artifact needs or contains reusable Skills.
Skill candidates include:
Tool procedure
Cognitive method
Review checklist
Extraction procedure
Transformation process
Evaluation protocol
Knowledge archival process
A Skill should have:
Trigger conditions
Input contract
Procedure
Output standard
Validation
Failure handling
Optional tools
11. Runtime Check
Determine whether Runtime is needed.
Runtime is needed if the artifact includes:
Multiple stages
Multiple agents
State tracking
Human decision gates
Tool or file operations
Handoff
Recovery
Archival
Report synthesis
If not needed, do not create Runtime.
12. Creation Brief
Before generating final artifacts, produce a Creation Brief.
Use this structure:
# Creation Brief
## 1. Intended Use
## 2. Target User
## 3. Target Platform
## 4. Artifact Classification
## 5. Operating Mode
## 6. Depth vs Automation Orientation
## 7. Cognitive Models Involved
## 8. Skills Needed
## 9. Runtime Need
## 10. Human Decision Gates
## 11. Proposed Files
## 12. Acceptance Criteria
13. Proposed Files
List target paths before writing.
Examples:
agents/lite/zhangliao-red-team.prompt.md
agents/agent-specs/cognitive-imaging-specialist.agent.md
skills/cognitive/cognitive-imaging.skill.md
model-cards/intermediate/cognitive-imaging-model.md
runtimes/interactive/modeling-committee.runtime.md
14. Artifact Generation Rules
When generating an artifact:
Use the correct template.
Keep the artifact fit-for-purpose.
Do not include unnecessary layers.
Preserve user terminology.
Mark human decision gates.
Include evaluation rules.
Include version metadata when durable.
Include platform notes when relevant.
15. Validation Checklist
Before finalizing, check:
Is the classification correct?
Is the objective clear?
Is the artifact over-engineered?
Is the artifact under-specified?
Are model assets separated when needed?
Are Skills extracted when useful?
Are human decision gates explicit?
Is the output format usable?
Is the target path appropriate?
16. Creator Mode Final Response
Final response should include:
Mode used: Creator Mode
Artifact classification
Operating mode
Depth vs automation orientation
Files generated or proposed
Human decisions needed
Recommended next action
17. Final Rule
Creator Mode should help the user build new AI artifacts without losing conceptual depth.
Create enough structure to make the artifact durable.
Do not create more structure than the artifact needs.