Self-Improvement Agent Harness: A Deterministic SIA Exemplar

Abstract This exemplar documents template sia, a deterministic implementation of the Self-Improvement Agent (SIA) harness contract described in . The default pipeline replays fixture-backed generations for the mini classify task; opt-in live mode runs bounded target subprocesses and optional Ollama-backed meta/feedback steps. Run snapshot. Task mini classify, run 1, 3 generation(s), live=false. Final accuracy=0.8333 over 6 held-out samples. Values are injected by scripts/z generate manuscript variables.py after analysis. Keywords: self-improvement agents, benchmark harness, reproducible evaluation, agent loops --- Associated artifacts GitHub release: v0.1.0 (https://github.com/docxology/template/releases/tag/v0.1.0) PDF SHA-256: 7087b6d1dd2e6192b24055408935201c27d7e19ba26267a366b20b2c71b7a721