skills.rest
thistleknot
thistleknotCOMMUNITY
Explore all skills by this author →
WebsiteGitHub

react_agent

Automates multi-step software tasks with a structured Plan-Execute-Validate workflow.

Updated Apr 20, 2026
Search Tags:#agentic workflow#autonomous coding#test execution#workspace memory#software refactoring#reason-act-observe#action validation
One-click install
npx skills add https://github.com/thistleknot/skills --skill react-agent-thistleknot
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: react_agent
Source: https://github.com/thistleknot/skills/tree/main/react-agent
Command: npx skills add https://github.com/thistleknot/skills --skill react-agent-thistleknot

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill turns complex, multi-step software work into a structured loop that plans changes, validates actions, observes results, and keeps execution on track until the work is actually done.

Core Features & Use Cases

  • Reframe → Recon → Plan → ReAct Execute → Completion: Converts an ambiguous request into an execution contract, then iterates through actions with explicit validation and completion checks.
  • Action Harness Validation (HARNESS): Prevents illegal or out-of-scope changes via structural, semantic, and environmental checks, with a typed failure taxonomy and retry/pivot protocol.
  • Persistent Working Memory (.react_agent/) + Retrieval (MEMORY): Stores task state, progress, evidence, and episodic cases to avoid repeating prior mistakes and to reuse what worked.
  • Quality Gate (QUALITY): Applies a six-dimension evaluation (Safety, Completeness, Executability, Maintainability, Cost-Awareness, Acceptance Alignment) before declaring success.

Quick Start

Use the react_agent skill to refactor a multi-module codebase by creating an execution plan, applying validated edits, running tests, and producing an evidence-backed completion report.

Frequently Asked Questions about react_agent

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

FAQPage Schema
How do I automate complex software refactoring tasks safely?▼

Automate complex software refactoring safely by using a structured Reason-Act-Observe loop that validates actions, runs tests, and checks explicit completion criteria before finalizing changes.

How does an autonomous coding agent validate actions before applying changes?▼

An autonomous coding agent validates actions through an action harness that performs structural, semantic, and environmental checks, preventing illegal or out-of-scope changes via a typed failure taxonomy.

Can I use an agentic workflow for multi-step code migrations and feature implementation?▼

Yes, you can use an agentic workflow for multi-step code migrations and feature implementation by converting requests into execution contracts and iterating through validated edits with evidence-backed completion.

What is the best way to ensure autonomous code execution meets acceptance criteria?▼

Ensure autonomous code execution meets acceptance criteria by applying a six-dimension quality gate that evaluates safety, completeness, executability, maintainability, cost-awareness, and acceptance alignment before declaring success.

Do I need a specific directory structure to run an autonomous coding agent?▼

Yes, autonomous coding agents require a persistent workspace state directory to store task state, progress, and evidence, enabling iterative refinement and memory-backed retrieval throughout the execution loop.

Related Skills

Explore More Software Engineering Skills→
bostonaholicbostonaholic
8

team

Orchestrate an autonomous multi-agent pipeline from feature description to pull request.

Community
Advanced
loriensleafsloriensleafs

memory-documentary

Query Brain memory, artifacts, GitHub issues to assemble cited documentary reports.

Community
Advanced
nguyenphpnguyenphp
41

lint-and-validate

Run linters, type checkers, and security audits across Node.js and Python projects.

Community
Intermediate
shawnroosshawnroos

auto-preset

Dispatch a named preset against a target with ephemeral verification criteria.

Community
Intermediate