clawflow

Manage long-lived owner-scoped jobs with persisted state and waiting across detached tasks.

Updated Aug 23, 2026
One-click install
npx skills add https://github.com/chiragborse1/Kove --skill clawflow
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: clawflow
Source: https://github.com/chiragborse1/Kove/tree/main/skills/clawflow
Command: npx skills add https://github.com/chiragborse1/Kove --skill clawflow

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

ClawFlow provides a lightweight runtime for jobs that must outlive a single prompt or detached run while preserving a single owner session, thread context, and a small persisted output bag so work can be waited on, resumed, inspected, and finished with clear ownership.

Core Features & Use Cases

  • Flow identity and ownership: Maintain a single owner session and flowId for multi-step background jobs.
  • Waiting and resumption: Pause flows while detached tasks or subagents complete, then resume with preserved state.
  • Small persisted outputs and lifecycle: Store brief outputs, emit owner-facing updates, and support finish, fail, cancel, and blocked states.
  • Use Case: Triage an inbox by creating a flow, run classification tasks in detached workers, set waiting state, then resume to route results and emit a single owner update.

Quick Start

Create a flow to manage a multi-step background job for a single owner, run tasks in the flow, set waiting or outputs, and resume or finish the flow as needed.

Frequently Asked Questions about clawflow

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I manage state for long-lived background workflows that outlive a single prompt?

Manage long-lived background workflows by using a runtime that preserves a single owner session, flow identity, and a small persisted output bag. This allows multi-step jobs to maintain state and thread context across detached runs.

What is the best way to pause a background task and resume it after a detached subagent completes?

The best way to pause a background task is to set a waiting state that preserves flow identity and ownership. Once the detached subagent or ACP run finishes, the flow resumes with its prior state intact.

Can I use flow control semantics like finish, fail, and cancel for owner-bound jobs?

Yes, you can apply finish, fail, and cancel semantics to owner-bound jobs. The runtime supports these lifecycle states along with blocked and waiting states, enabling clear ownership and flow control without implementing business branching logic.

Does this workflow orchestration approach work for multi-step jobs requiring task linkage?

Yes, workflow orchestration works for multi-step jobs requiring task linkage. It provides a single owner session, links detached tasks or subagents, and persists small outputs so work can be inspected, resumed, and finished.

When do I need resumable jobs with waiting state for background tasks?

You need resumable jobs with waiting state when multi-step background tasks must pause for detached workers to complete. This preserves ownership and persisted outputs, allowing the main flow to resume and route results seamlessly.

Why does my multi-step flow lose context after a detached ACP run finishes?

Your multi-step flow loses context because it lacks a persistent owner session and flow identity. By persisting small outputs and maintaining a waiting state, the runtime preserves thread context across detached ACP runs.