workflowx

Execute Mode B workflow (local module, PRD detection)

Optimize prompt without triggering workflow

Invoke the orchestratorX agent in status mode to scan the project, collect workflow state, and generate an HTML status report.

Execute Mode C workflow (unit task, minimal change)

Execute Mode A workflow (whole codebase, full planning phase)

Lean code & engineering analysis agent. Owns structured Markdown code summarization and must apply abstracter-code-summary skill for detailed analysis workflow.

Code implementation teammate. Works in Agent Teams mode, communicates directly with evaluator-teammate. Claims coding tasks, implements code, marks completion.

Lean coding agent. Owns implementation of features, bug fixes, refactors, and tests. Must apply karpathy-guidelines skill as the behavioral baseline for all coding work. Supports Hybrid Tree workflow with Parent+Child document reading and maintenance.

Code evaluation teammate. Works in Agent Teams mode, communicates directly with coder-teammate. Reviews code, generates evaluation reports, discusses fix plans.

Lean code audit and evaluation agent. Reads Parent+Child hybrid docs for ground truth, inspects git diffs and project code, then produces structured evaluation reports in Hybrid Tree workflows.

Code & engineering analysis skill for abstracter agents. Use this skill whenever the user asks to summarize code, analyze project structure, review a module or subsystem, produce a structured Markdown analysis report, or evaluate code quality, risks, and improvement suggestions — even if they only provide a file path or directory name.

Intelligent request classification and routing. Status gate, 4 routes (active workflow, direct handling, workflow selection, explicit command), 5-dimension analysis, mode recommendation, status management. Used by main Claude agent before dispatching orchestratorX.

Spec-driven coding workflow for coderX agents. Use this skill whenever implementing, fixing, refactoring, or iterating code from hybrid docs or specified requirements, especially when file index, knowledge index, or evaluator findings are present.

Structured code audit workflow supporting both PRD-based and prompt-based evaluation modes. evaluatorX is a pure analyzer — reads docs + code, outputs structured Evaluation Result Payload. Document writes are handled by orchestratorX, not evaluatorX.

Behavioral guidelines to reduce common LLM coding mistakes. Use when writing, reviewing, or refactoring code to avoid overcomplication, make surgical changes, surface assumptions, and define verifiable success criteria.