until-done - Task execution loop with verification
Self-referential loop until task completion with architect verification
Tags
Updated: 2026-05-09Capabilities
Typical Inputs
Typical Outputs
What this skill does
- delegate tasks to agents
- execute tasks in parallel
- run background operations
- run verification tests
- read verification output
- spawn architect verification
- cancel execution
Inputs
- original task requirements
- TODO list
- iteration state
- verification commands
- model parameter
Outputs
- completed task
- verification results
- test outputs
- build outputs
- architect approval
Requirements
- agent tiers support
- architect role availability
- verification tools
- state persistence