Problem
When debugging micro-agent behavior, there's no way to preview what test prompt will be sent to the LLM without actually running it.
Proposed Solution
Add a (or ) flag that outputs the generated test prompt to stdout and exits without making any LLM calls or file changes.
Example:
micro-agent ./utils.ts -t "npm test" --dry-run
# Outputs generated test prompt to stdout, no execution
Use Case
Useful for debugging, verifying prompt engineering, and teaching how micro-agent works before committing to execution.
Problem
When debugging micro-agent behavior, there's no way to preview what test prompt will be sent to the LLM without actually running it.
Proposed Solution
Add a (or ) flag that outputs the generated test prompt to stdout and exits without making any LLM calls or file changes.
Example:
Use Case
Useful for debugging, verifying prompt engineering, and teaching how micro-agent works before committing to execution.