agentic-harness

What problem does it solve?

agentic-harness solves the problem of unstable, illegal, or incoherent outputs produced by autonomous LLM coding pipelines by turning them into a legality-checked, state-tracked, self-repairing workflow.

Core Features & Use Cases

• Coherence-gated execution loop: uses an explicit coherence todo flag and only marks completion after a representative run actually passes.
• State machine over conversation: replaces chatty “keep going” behavior with validated transitions and tool-checked gates.
• Structured response wire format: enforces typed payloads at every machine-to-machine boundary to prevent prose-driven truncation and ambiguity.
• Critic agent ensemble option: for difficult tasks, runs conservative/balanced/creative critics and aggregates blocking issues into an ACCEPT/REVISE/ESCALATE decision.
• Anti–band-aid repair discipline: fixes the mechanism that produced failure classes, validated by minimal repro + full harness run.

Quick Start

Tell an AI to use agentic-harness to run a bounded coding pipeline, enforce legality and coherence gates, and only finalize when the expected artifacts reopen and verification passes.