Tag: agent workflows

This is valuable as a concrete account of making models replaceable. Shared skills and repository instructions create a common working surface; frontier models handle exploration or difficult planning; cheaper models execute established patterns; critics challenge the result; and scripts preserve repeatable work without spending tokens again. The harness becomes the place where the developer's standards, tools, and accumulated workflow live, so switching providers does not mean rebuilding the practice from zero.

It is not a universal blueprint. Model routing, role handoffs, large plans, and multiple subscriptions can save money in one workload while adding context loss, coordination cost, and another system to debug in another. The best evidence is not the elegance of the orchestration diagram but the defects, review time, total cost, and maintainability it produces over repeated work. A good harness keeps the human's mental model intact, makes outputs auditable, and lets the team change models without surrendering its engineering process. When the workflow starts guiding the developer more than the developer guides the work, it is time to simplify it.

This is more useful than another argument about vibe coding because it is a concrete trace: notes, mockups, data modeling, scaffolding, implementation transcripts, and maintenance around Diggit, a small git-history tool for architecture archaeologists.

The point I want to keep is that good agent work still starts before the agent. Taylor narrows the problem with plaintext design, simple Elm scaffolding, explicit model shapes, fast feedback, and loud states. Claude wrote most of the code, but the product is really the constraint system around Claude.

This overlaps with the coding-assistant posts on the site, but it is still worth keeping as an outside practitioner account: small steps, screenshots, AGENTS.md, undo discipline, and refactoring agent-generated blobs.

The caveat is that advice like this should become process, not vibes. If an assistant can break the project quickly, the workflow has to make rollback and review cheap.