Coordinate recurring shift execution for long-running AI workflows.
npx skills add https://github.com/EverClaw/EverClaw --skill three-shifts-everclaw
Please help me install this Agent Skill. Skill: three-shifts Source: https://github.com/EverClaw/EverClaw/tree/main/three-shifts Command: npx skills add https://github.com/EverClaw/EverClaw --skill three-shifts-everclaw
This Skill turns a long-running workflow into a reliable shift-based execution system, so approved work is planned once, broken into small steps, and completed without losing progress to timeouts or crashes.
Use the three-shifts skill to read the current shift files, decompose the approved work into atomic steps, and update the shift state for the next cycle.
High-intent search queries and answers about installing and using this skill.
Shift cycle execution breaks approved work into small atomic steps processed in 15-minute loops. It requires persistent state files to track task status and uses cron-driven scheduling to complete steps safely without losing progress.
Stateful handoffs persist tasks, context, and results across runs using persistent state files. The next cycle or shift reads these files to continue cleanly, automatically managing carryover items and approval-aware task selection.
Stepwise task decomposition breaks an approved daily work queue into one small step per 15-minute execution loop. Each atomic step is marked as complete, blocked, skipped, or in progress to ensure safe execution.
Yes, persistent state files are required to store shift plans, carryover items, and execution results. The cron-driven scheduling relies on these files to maintain state across morning, afternoon, and night planning cycles.
Approval-aware planning requires an operator to approve a daily work queue before execution begins. The system then decomposes the approved tasks into safe steps and executes them, relying on retry-safe error logging for failures.
Workflows timeout when processing too much work in a single run. Shift cycle execution solves this by breaking work into atomic 15-minute cycles with end-of-shift handoff generation, allowing tasks to resume in the next cycle.
profbernardojPlan overnight autonomous work with approval-gated task execution for EverClaw.
HwFeeParse user report requests into a structured JSON requirements schema.
samymityGuide agentic engineering workflows with eval-first execution and model-tier routing.
profbernardojDecomposes high-level tasks into atomic steps and executes them in 15-minute cycles with persistent state.
