task-flow-handoff

Writes session handover records when context window fill interrupts ongoing work.

1|Updated Jan 14, 2024
One-click install
npx skills add https://github.com/Eyhenij/rt-tools --skill task-flow-handoff-eyhenij
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: task-flow-handoff
Source: https://github.com/Eyhenij/rt-tools/tree/main/.claude/skills/task-flow-handoff
Command: npx skills add https://github.com/Eyhenij/rt-tools --skill task-flow-handoff-eyhenij

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? When a long AI coding session hits context window limits or is interrupted, uncommitted progress and session state are lost, forcing the next session to redo work or guess where things stand. This Skill defines how to close a session cleanly: pick a stopping point, record progress, commit verified work, and write a structured handover. ## Core Features & Use Cases - Threshold-based closing: Defines behavior at window-fill thresholds — pick a stopping point at the first reminder, refuse everything but the handover at the second. - Structured handover format: Specifies fixed sections (Work, Where to look, Epic position table, Done and next step, What to keep in mind) written into the task's progress file and committed to the branch. - Hook integration: A hook writes a handover draft before context compaction and injects it into the next session at startup, so no manual copy-paste is needed. - Use Case: A coding session reaches 80% context fill mid-task. The Skill directs the executor to finish the current plan stage, commit verified work, write the handover section with branch, PR number, and next step, and name the handover path to the owner. ## Quick Start Close this session following the task-flow-handoff pattern: record progress, commit verified work, and write the handover section for the next session.

Frequently Asked Questions about task-flow-handoff

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

FAQPage Schema
How do I hand off an unfinished coding task between AI sessions?

Write a handover section into the task's progress file with fixed sections: Work (branch, tree path, PR), Done and next step, and session-specific notes. Commit and push it so the next session picks up exactly where work stopped.

What should a session handover document contain?

A handover contains the task number and branch, what is done and the next step, an epic position table if applicable, and particulars not in the rules or progress such as raised stands, lagging dependencies, or background reviews awaiting results.

When should a session start closing on context window fill?

Closing starts at the first fill threshold reminder, not at refusal. The session picks a logical stopping point, brings the current step to it, commits verified work, and writes the handover before the second threshold blocks everything else.

Can a handover survive moving work to another machine?

Yes. The handover section lives in the task's progress file, which is committed and pushed to the branch. Work without a task folder falls back to a standalone handover file in a profile-named directory.

Why should handover state not include things like build freshness?

States that change without the executor, such as lag behind main or check results, go stale within hours and mislead the next session. The handover records only stable facts: the tree, branch, PR number, and uncommitted work.