skill-orchestrator

Pipeline construction declares parallel, sequential, or hybrid phases to coordinate multi-concern engineering tasks. It injects resolved stack context into agent prompts and enforces review gates between phases to prevent conflicting outputs across UI, API, and database domains.

Stack detection resolves technology context from cache or fresh detection based on staleness rules. The resolved stack context is then injected into agent prompts to ensure cross-cutting refactors target the correct frameworks and architectural patterns.

Yes, architectural migrations are supported by classifying complexity and mapping affected domains to primary and auxiliary skills. The orchestrator builds a phase-based pipeline with mandatory code-reviewer inclusion to ensure migration phases are sequenced and verified.

The best way to prevent coordination failures is explicit skill assignment with mandated code-reviewer inclusion. The orchestrator resolves skill conflicts through prescribed precedence and writes detailed results to an append-only log, ensuring every concern is reviewed and sequenced.

You need an append-only orchestration log to capture detailed execution results and skill gaps for continuous improvement. This self-logging mechanism provides an immutable history of pipeline phases, enabling analytics to refine future multi-agent coordination and task classification.