How to maintain agent-built codebases: a weekly frontier-model architecture audit, one improvement at a time
vikvang1 · x · 2026-11-10
Engineer BHolmesDev poses a question: when everything from hello world onward is agent-built, how do you keep the codebase navigable for both agents and humans?
His answer is system architecture — agents copy bad patterns like a virus, so the codebase is a ball of clay needing constant reshaping. Writing best practices into your factory's skills helps, but keeping the codebase on the right path is the real goal. Having an agent review everything and spit out architecture PRs nobody reads is too expensive; a workable start is a weekly automation with a frontier reasoning model (he suggests Fable 5.2) that pulls out one system-architecture improvement at a time, borrowing @steveruizok's prompt to "look for implicit state machines." He plans to experiment with this factory automation.
More from coding & agent
- Fable 5.2 vs Opus 5.5: default to Opus, upgrade to Fable for long builds and messy context — every · 2026-11-10
- Neo4j open-sources an ontology spec: versioned YAML schemas with JSON Schema validation for GraphRAG — JeremyCMorgan · 2026-11-10
- Gmail MCP Server adds a one-tool send_email agent capability via your Gmail credentials — modelcontextprotocol · 2026-11-10
- HubSpot MCP Pack connects agents to HubSpot CRM over Streamable HTTP — modelcontextprotocol · 2026-11-10
- SlopOps: mistaking agentic coding capability for competitive advantage is playing the slop lottery — generativist · 2026-11-10
- Dev kicks off building an agentic game engine in Rust powered by Fable 5.2 — EricBuess · 2026-11-10