agent

Automate local code review and edits with an Ollama agent.

2|Updated Apr 4, 2026

Search Tags:#agent #memory #code-review #tool-calling #git-safety #local-automation

Ready to run

npx skills add https://github.com/Nickalus12/Loom --skill agent-nickalus12

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: agent
Source: https://github.com/Nickalus12/Loom/tree/main/claude/skills/agent
Command: npx skills add https://github.com/Nickalus12/Loom --skill agent-nickalus12

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automates local code review and edits by running an Ollama agent on your machine, enabling tool-calling to read, modify, and analyze code without manual context switching.

Core Features & Use Cases

Dual-model routing for fast tool calls and in-depth analysis
7 tools: read_file, read_file_lines, edit_file, write_file, search_code, find_files, run_powershell
Git safety: auto-create a dedicated branch before writes and show diffs
Think-then-act, caching, and syntax validation to improve reliability
Session memory via Graphiti to preserve task context across steps

Quick Start

Use the local_agent_task MCP tool with your task description to start an autonomous coding review.

Frequently Asked Questions about agent

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

FAQPage Schema

How do I automate local code review and editing on my workstation?▼

Automate local code review and editing by running an Ollama agent that uses tool-calling to read, modify, and analyze project files without manual context switching. It applies caching, syntax validation, and session memory to ensure deterministic code modifications.

What is the best way to ensure Git safety during automated code modifications?▼

Ensure Git safety during automated modifications by using a workflow that auto-creates a dedicated branch before writing files and shows diffs. This approach prevents direct commits to your main branch and provides auditable results for every change.

Does this local automation approach require an Ollama agent to run?▼

Yes, local automation requires running an Ollama agent on your machine. The agent implements dual-model routing for fast tool calls and in-depth analysis, using seven integrated tools to read, edit, search, and execute commands.

Can I preserve task context across multiple steps of a code analysis workflow?▼

Preserve task context across multiple code analysis steps using session memory backed by Graphiti. This memory integration allows the agent to maintain state and reference previous actions throughout the entire review and editing process.

What tools are available for autonomous code analysis and modification?▼

Autonomous code analysis and modification tools include read_file, read_file_lines, edit_file, write_file, search_code, find_files, and run_powershell. These seven tools enable comprehensive file manipulation, codebase searching, and local command execution.

When should I not use an autonomous agent for local code edits?▼

Avoid using an autonomous agent for local code edits when you lack a local Ollama setup or need strict manual oversight without automated branching. The agent autonomously applies changes and creates branches, which requires a compatible local environment.