kanban-worker

Reliable Kanban worker handoffs require using kanban_show before starting, then shaping kanban_complete, kanban_block, and comment payloads with structured metadata and review-required reasons. This prevents low-signal outcomes from reaching downstream human reviewers.

Workspace isolation for multi-tenant Kanban workers ensures correct behavior across scratch directories, persistent shared state, and git worktrees. This prevents stale artifacts and improper file modifications across isolated tenant memory boundaries during worker execution.

You need to block a Kanban task for human review when outcomes require manual validation before proceeding. Safe review gating uses kanban_block payloads to provide downstream tools and humans with structured reasons to decide next actions reliably.

Diagnosing retry scenarios for failed Kanban tasks involves interpreting prior run outcomes to avoid repeating failing paths. Safe retry hygiene requires reviewing previous execution metadata to identify and bypass previously unsuccessful worker execution routes.

The best way to structure kanban_complete payloads is to include structured metadata, outcomes, and review-required reasons. This standardization allows downstream tools and human reviewers to reliably parse worker results and decide next actions without ambiguity.

Claiming nonexistent created cards in Kanban workflows breaks downstream parsing and misleads human reviewers. Correct completion semantics require providing accurate metadata and outcomes without fabricating card creation to maintain workflow handoff reliability.