Prompt Maximizer Skill

Purpose

This skill transforms any user prompt or task request into a clearer, more complete, more executable prompt before the agent acts on it.

Use this skill as a pre-execution gate.

The goal is not to make prompts longer for their own sake. The goal is to make them:

• clearer
• safer
• more specific
• more complete
• better scoped
• easier for an agent to execute
• less ambiguous
• easier to verify

When to Use

Use this skill whenever the user provides:

• a prompt they want improved
• a task they want the agent to execute
• an instruction for another agent
• a system prompt
• a coding request
• a research request
• an automation request
• a command-generation request
• a troubleshooting request
• a content-generation request
• a planning request
• any vague or underspecified instruction

Default assumption: if the user gives a prompt or asks the agent to do something, this skill should run first.

Do Not Use When

Do not run the full verbose optimization process when:

• the user explicitly says: "do not improve the prompt"
• the user asks for a very small direct answer
• optimizing would delay an urgent safety-critical answer
• the user is asking for a simple factual answer and no prompt will be reused
• the user has already provided a strict final prompt and only wants execution

Even then, silently apply light clarity improvements when useful.

Core Workflow

For every eligible prompt, follow this five-step workflow.

Step 1: Capture the Raw Prompt

Identify the user's actual instruction.

Preserve:

• original goal
• constraints
• tone
• language
• target model or agent
• tools mentioned
• security boundaries
• output format
• success criteria

If something is missing, infer reasonable defaults instead of asking too many questions.

Step 2: Diagnose Prompt Weaknesses

Check for:

• vague goal
• missing context
• missing constraints
• missing output format
• missing role
• missing examples
• missing success criteria
• hidden assumptions
• unsafe or overly broad tool permissions
• unclear order of operations
• lack of verification step
• lack of stop conditions
• lack of error handling
• no distinction between planning and execution

Step 3: Produce an Optimized Prompt

Rewrite the prompt using this structure:

1. Role
2. Mission
3. Context
4. Inputs
5. Constraints
6. Allowed actions
7. Forbidden actions
8. Process
9. Verification
10. Output format
11. Escalation / clarification rules

The improved prompt should be directly usable by another agent.

Step 4: Self-Critique the Optimized Prompt

Ask internally:

• Is the goal explicit?
• Are tool permissions clear?
• Are dangerous actions gated?
• Does the prompt prevent unnecessary verbosity?
• Does it include verification?
• Does it define output format?
• Does it handle ambiguity?
• Does it reduce hallucination risk?
• Does it preserve the user's intent?

Then improve the prompt one more time.

Step 5: Output or Execute

If the user asked for a prompt:

• show the optimized prompt
• show a short explanation of what changed

If the user asked the agent to perform the task:

• use the optimized prompt internally
• briefly mention that the request was normalized
• proceed with the task
• do not dump the full optimized prompt unless useful

Mandatory Output Format When User Asks For A Prompt

Use this format:

# Optimized Prompt

[final optimized prompt]

# Why This Is Better

- [short explanation]
- [short explanation]
- [short explanation]

# Optional Compact Version

[shorter version if useful]