config-tuning

Configuration tuning analyzes patterns in opencode.jsonc, AGENTS.md, and skill definition files to identify issues affecting agent behavior. It proposes concrete, structured changes with tradeoffs and requires your approval before applying edits, reducing workflow friction and improving performance.

Pattern recognition examines your current config files to surface improvement opportunities in agent behavior and workflow efficiency. Common patterns include redundant settings, misaligned skill dependencies, and suboptimal command definitions that slow iteration or create agent confusion.

Yes. Configuration tuning analyzes your existing config to identify where new workflows or skills fit naturally into your stack. It proposes structured additions with rationale, helping you integrate changes without breaking current agent behavior or workflow dependencies.

Configuration tuning automates pattern analysis and proposes option-by-option changes with documented tradeoffs, rather than making edits blindly. It surfaces best practices and integration points you might miss, then guides you to OpenSpec or targeted tweaks for faster, safer iteration.

Configuration tuning examines opencode.jsonc for overall workflow structure, AGENTS.md for agent behavior definitions, and skill/command/definition files for granular settings. This multi-file analysis reveals how config choices interact and where friction points exist across your entire setup.

No special tools are required. You need access to your opencode.jsonc, AGENTS.md, and skill definition files. Configuration tuning guides you through pattern analysis and explains proposed changes clearly, so you can make informed approval decisions without deep config expertise.