AuricIDE
Blog
Notes from building AuricIDE: agent orchestration, machine-checkable goals, and what it actually takes to stop babysitting AI agents.
-
AI Development Agents: Why the Config Layer Matters Most
AI development agents depend less on the model and more on the config and permission layer around it. See how dynamic provider configs actually work.
-
Open Source AI Tools for Developers in 2026
Curated open source AI tools for developers in 2026, from agent CLIs and local models to MCP servers, with trade-offs and clear use cases.
-
LLM Agent Architectures: Three Shapes That Matter
A working engineer's guide to LLM agent architectures, the three shapes that define them, and the trade-offs each one carries
-
How to Create an AI Agent That Actually Ships
Learn how to create an AI agent that goes from a single CLI tool to a coordinated fleet, with practical steps for architecture, skills, and state.
-
How Much Does an AI Agent Cost: The Real Structure of Spend
How much does an AI agent cost? Skip the outdated price lists. Learn the actual structure of agent spend, from token loops to orchestration overhead.
-
AI Agent Access Control in AuricIDE
Learn how AI agent access control works in AuricIDE. Discover permission modes, MCP enforcement, and practical strategies for securing local coding agent
-
AI Agent vs LLM: When to Use Each for Engineering Workflows
AI agent vs LLM explained for engineers. Learn the architectural differences, when multi-step orchestration pays off, and how to choose the right approach.
-
AI Agent Monitoring: Why Exit Codes Aren't Enough
Each monitoring signal proves less than its label implies. What process state, exit codes and the repository diff can each tell you, and what none of them can.
-
AI Agent Orchestration Framework: What It Actually Has to Solve
A second agent does not double the work, it changes it. What an orchestrator has to control, and when a shell script is still the better answer.
-
AI Agent Testing: Assert Behavior, Not Wording
A red agent test can point at the check rather than the agent. What to assert instead: exit codes, stored project state, and where a chain stops.
-
MCP Server Setup: Why a Running Server Proves Nothing
A running MCP server only proves the launch worked. How to write the launch entry, why a wrong database path still lists every tool, and the test that catches i
-
What Is Agentic Coding and How Does It Actually Work
Learn what is agentic coding, how it differs from autocomplete, and how multi-agent orchestration works in practice through concrete examples.