Review agent runtime code for architectural violations and separation of concerns. Enforces layer boundaries between inference (ollama-client), runtime (agent-runtime), orchestration (agent-orchestrator), and framework (ares).
Installation
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Review agent runtime code for architectural violations and separation of concerns. Enforces layer boundaries between inference (ollama-client), runtime (agent-runtime), orchestration (agent-orchestrator), and framework (ares).
disable-model-invocation
true
argument-hint
["file-or-directory"]
allowed-tools
Read, Glob, Grep
Skill: Runtime Architecture Check
Review agent runtime code for architectural violations and separation of concerns.
Rules
Inference layer (ollama-client) must not contain orchestration logic
Orchestration layer (agent-orchestrator) must not call inference directly — go through agent-runtime
ares is the reference framework — custom agents must extend it, not reimplement primitives
No hardcoded model names — all model references must be runtime configuration
Agent state must be serializable (no closures or open file handles in state)
Skills must be stateless — they take input, return output, no globals
Output
List of layer boundary violations with file:line
Dependency graph of actual vs intended architecture