Symphony's README mentions it works best in codebases that have adopted harness engineering. I've been exploring this idea from the other direction — building a lightweight CLI tool called Project-Bootstrap-Harness (PBH) that helps developers quickly establish that environment.

What it does: pip install harness-init seeds a Python project with AGENTS.md (AI collaboration protocol), make verify (lint + test + coverage gate), and progress.json (phase tracking). The goal: any AI coding agent, regardless of tool or model, can understand the project within 30 seconds.

Experimental data: To test whether this actually helps, I ran a controlled comparison — same AI agent (Trae Solo Coder), same task (add a greet CLI subcommand with tests and docs), two starting conditions: an empty folder vs. a PBH-scaffolded project.

Key results:

Task steps reduced by 40% (10 → 6)

Debugging iterations reduced by 75% (4 → 1)

Total time reduced by 60% (15-20 min → 5-8 min)

New tests doubled (4 → 8)

Quality gate automatically executed in the PBH project (make verify), not in the bare one

Full experiment summary: summary.en.md

Honest caveat: This is a single data point. PBH provides a bundle of scaffolds rather than one isolated variable, so the experiment measures the combined effect of having a structured environment versus having none at all. I'd love to hear if others have done or would be interested in similar comparisons.

PBH repo: https://github.com/renjianguojinqianfan/Project-Bootstrap-Harness

Curious to hear the community's thoughts — and whether a quick-start harness engineering setup is something Symphony users would find useful.