What I Learned Using Free AI Coding Agents on a Real Portfolio (OpenCode and Kilo)

Description and use

Using free AI coding agents in VS Code sped up portfolio development but caused errors and delays. The author established a step‑by‑step process: read‑only diagnosis, small changes, manual review, local build, and deployment checks, reducing mistakes.

Problem

Relying on free AI coding agents in VS Code caused issues such as incorrect fixes, stalled execution, accidental commits, and environment mismatches.

Outcome

Adopting a disciplined, step‑by‑step workflow with AI agents reduced errors, prevented accidental commits, and ensured reliable builds and deployments.

Workflow

• Ask for a read‑only diagnosis
• Ask for one small change
• Run git status and git diff --stat
• Run the build locally
• Test the page on desktop and phone-sized window
• Commit with a clear message and push
• Check deployment status is Ready
• Verify live site in incognito tab

Evidence

They saved me a lot of time, but they also got stuck, lied by accident, and once looked busy for a whole hour without doing anything.

Source

What worked was splitting the job into two steps: First, read only. Do not change anything. Find where the mobile menu is rendered and tell me exactly why the panel is invisible, with the file and line number. Then fix it.

Source

Short prompts with one job each worked much better.

Source

Agents are happy to "help" by committing and pushing. I wanted to review every change first.

Source

Typecheck passing does not mean the build passes

Source