How does the exploration phase work?

It uses multiple specialized code-explorer agents to map architecture, trace execution paths, and identify similar patterns before any code is written.

How does Finesse decide when to decompose a task?

It evaluates specific metrics such as estimated iterations (>25), number of files touched, and functional independence to propose splitting a task into sequenced sub-loops.

Can I skip the UAT checkpoints?

Yes, users can choose to fast-forward UAT at any checkpoint to proceed directly through the remaining phases of the workflow.

What task types are supported by this skill?

It includes dedicated workflows for features, bug fixes, refactors, testing, performance optimization, and research tasks.

Finesse Task Workflows

Name: Finesse Task Workflows
Author: jonathanung

byjonathanung

•

生産性とワークフロー

Standardizes AI-driven development cycles through task-specific planning phases and intelligent scope decomposition.

Finesse Task Workflows provides a structured framework for Claude Code to handle complex engineering tasks by categorizing them into specialized workflows like features, bug fixes, and refactors. It moves beyond simple prompting by introducing deep discovery phases, architectural design multi-agent reviews, and intelligent scope decomposition for tasks that exceed standard iteration limits. With integrated User Acceptance Testing (UAT) checkpoints and automated exploration caching, it ensures that Claude operates with maximum context and precision throughout the entire development lifecycle.

主な機能

01Intelligent Decomposition: Analyzes scope breadth to split massive tasks into manageable, sequenced sub-workflows.

02Integrated UAT Checkpoints: Enforces human-in-the-loop validation at critical architecture and planning stages.

03Multi-Agent Architecture Design: Orchestrates specialized agents to propose and validate diverse implementation strategies.

044 GitHub stars

05Automated Context Caching: Persists codebase exploration findings to reduce redundant analysis and optimize token usage.

06Task Type Classification: Automatically detects and applies optimized workflows for bugs, features, and refactors.

ユースケース

01Implementing complex new features that touch multiple modules and require architectural oversight.

02Executing large-scale refactors by decomposing them into a series of independent, safe iterations.

03Debugging deep-rooted regressions that require systematic tracing and reproduction paths.

主な機能

01Intelligent Decomposition: Analyzes scope breadth to split massive tasks into manageable, sequenced sub-workflows.

02Integrated UAT Checkpoints: Enforces human-in-the-loop validation at critical architecture and planning stages.

03Multi-Agent Architecture Design: Orchestrates specialized agents to propose and validate diverse implementation strategies.

044 GitHub stars

05Automated Context Caching: Persists codebase exploration findings to reduce redundant analysis and optimize token usage.

06Task Type Classification: Automatically detects and applies optimized workflows for bugs, features, and refactors.

ユースケース

01Implementing complex new features that touch multiple modules and require architectural oversight.

02Executing large-scale refactors by decomposing them into a series of independent, safe iterations.

03Debugging deep-rooted regressions that require systematic tracing and reproduction paths.