deepwork

Orchestrates phased multi-specialist coding workflows with Oracle review gates and persistent progress state.

Updated May 23, 2021
One-click install
npx skills add https://github.com/Favot/.dotfiles --skill deepwork-favot
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: deepwork
Source: https://github.com/Favot/.dotfiles/tree/main/.config/opencode/skills/deepwork
Command: npx skills add https://github.com/Favot/.dotfiles --skill deepwork-favot

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Large, high-risk coding efforts with multiple dependent phases often fail due to lost context, unreviewed intermediate work, and poor coordination between specialist agents. This Skill structures those efforts into planned phases with review gates and a persistent progress file. ## Core Features & Use Cases - Phased Orchestration: Splits large work into coherent implementation phases based on dependencies and delivery boundaries, delegating to specialist agents like @designer, @fixer, and @librarian. - Oracle Review Gates: Runs an @oracle review after each planned phase, with batched remediation passes and follow-up reviews only when decisions or risks change. - Persistent Session State: Maintains a markdown progress file under .slim/deepwork/ capturing research findings, plans, review notes, and validation results, kept git-local via .gitignore and .ignore rules. - Use Case: When migrating a cross-cutting architecture that cannot be partially shipped, activate deepwork to plan phases, delegate specialists, gate each phase with Oracle reviews, and track all decisions in one progress file. ## Quick Start Use the deepwork skill to plan and orchestrate this large multi-phase migration with specialist delegation and Oracle review gates after each phase.

Frequently Asked Questions about deepwork

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

FAQPage Schema
How do I orchestrate a large multi-phase coding task with AI agents?▼

Activate deepwork to draft a plan, split work into coherent phases based on dependencies, delegate each phase to specialist agents, and gate every phase with an Oracle review. Progress is tracked in a persistent markdown file under .slim/deepwork/.

When should I use deepwork instead of handling a task directly?▼

Use deepwork only for large or high-risk work: multiple dependent phases, cross-cutting architectural change, or migrations unsafe to partially ship. Do not use it for trivial edits, simple bug fixes, quick docs changes, or routine bounded features.

How do Oracle review gates work between implementation phases?▼

After each planned phase, the orchestrator validates results, updates the deepwork file with confirmed research context, and asks @oracle to review before continuing. Actionable findings are batched into one bounded remediation pass, with follow-up reviews only if decisions or risks change.

Does deepwork work with designer and fixer specialist agents?▼

Yes. When a phase includes @designer, the delivered UI/UX is treated as accepted design intent and preserved across later phases. @fixer is used only for bounded mechanical follow-up like wiring, tests, or type fixes that do not alter the design.

Where does deepwork store its progress and state files?▼

Deepwork creates a task-specific markdown file at .slim/deepwork/<task-slug>.md. The .gitignore is updated to exclude .slim/deepwork/ while .ignore whitelists it, keeping state git-local but readable by OpenCode.