Specification-driven development toolkit for DeepSeek Harness.
Recause is an experimental programming model and runtime for writing interactive software as ordinary code. Flows stop when information is missing, represent progress in explicit serializable state, and run again when new state arrives—without callbacks, explicit resume logic, or hand-authored state machines.
Local, resumable AI pipelines. A little TypeScript and a SQLite database.
Protocol for durable, resumable, evidence-backed work across long-running agents, humans, and tools.
Durable manual-core workflow skill for recursive planning, review/fix loops, and resumable AI-agent execution.
Agent skill for building local, resumable AI pipelines.
Clean-room Python reference for approval-gated, resumable bulk-file operations with synthetic fixtures and duplicate-aware tests.
Run record, resume decision, and reuse-history primitives for FlowBrigade-style workflows.
Agent-native workflow for translating full books with durable state, source-grounded review, resumable progress, and multi-book workspaces. Works with ChatGPT, Codex, and other AI agents.
To associate your repository with the resumable-workflows topic, visit your repo's landing page and select "manage topics."