sandeco-loop - Specify agent loops with checks and stop states
Conducts a short interview and writes a Loop Library specification for an agent loop without executing it.
Tags
Updated: 2026-09-29Capabilities
Typical Inputs
Typical Outputs
What this skill does
- Conduct a short interview
- Write a Loop Library document
- Define external stopping checks
- Add named terminal states
- Choose /goal or Ralph
- Ask for save location
- Create an optional slash command
Inputs
- Loop goal
- Success criteria
- Verification command or measure
- Trigger type
- Terminal states
- Named skills
- Sub-loop names
- Persistent memory location
- Guardrails
- Save location
- Slash-command preference
Outputs
- <name>-loop.md specification
- Optional loop-<name>.md command
- Interview questions
- Scheduled prompt when no loop is needed
Requirements
- Write access to the selected directory