go

The go skill acts as a sovereign, self-reflective execution loop specifically designed for the indiiOS AI-native operating system. It serves as an orchestration hub that ensures complex development tasks, ranging from codebase updates to deployment workflows, are driven to a verified, shipping state without manual intervention. By leveraging a structured approach to state snapshots, context scanning, and recursive re-evaluation, it minimizes agent drift and prevents the common stagnation found in multi-step AI workflows. It is intended for software engineers and creative developers working within the indiiOS repository who need a reliable mechanism to automate long-running implementation plans or complex multi-module updates.

• Automated recursive task management and loop-based execution for continuous progress.

• Built-in validation suite including type checking, linting, unit testing, and visual regression testing via browser subagents.

• Mandatory error memory checks using error ledgers and retrieval-augmented memory systems to prevent repetitive failures.

• Structured state management that tracks goal progress, session time, and dependency requirements.

• Rigorous commit cadence and git integration to ensure a clean working tree and maintain clear project history.

• Final 'Gauntlet' verification process requiring zero-error builds before tasks are marked complete.

• Invoke using /go or @go.md to initiate the loop; always ensure the session has an associated task.md and implementation_plan.md.

• The engine expects a deterministic project structure and relies on standard Node.js, TypeScript, and Electron development practices.

• Before execution, ensure the environment is primed with make prime and local secrets are configured via .env files.

• If a task fails more than twice, the agent will trigger a manual re-diagnosis rule to prevent the compound error trap.

• Ideal for managing dependencies between modules, verifying UI state, and maintaining consistency in complex multi-agent environments.

• Use the provided completion summaries to audit task N of M status, build outcomes, and visual verification evidence.