Delivery Harness

Delivery Harness is an educational reference implementation for on-demand delivery workflows, combining deterministic timelines, rules, and model endpoints.

Delivery Harness is an educational platform designed for on-demand delivery operations. It combines deterministic workflows, mock business tools, lightweight knowledge retrieval, an OpenAI-compatible model endpoint, evaluation scaffolding, and human-review guardrails. The system implements two synchronous workflows for abnormal-order analysis and compensation suggestion, using locally registered tools for orders, ETA, station capacity, and compensation rules. Compensation amounts are decided entirely by the rule engine rather than the model. It features in-memory document ingestion, lexical retrieval, request trace IDs, per-step durations, and a local Ollama profile. The platform is built as a Spring Boot application using Java 17 and Maven.

Delivery Harness at a glance

Best for: Software engineer
Runs on: API, CLI, self-hosted, open source
Pricing: Free
Released: August 2026
Category: AI agent tool
Website: github.com/diudiu-tech/delivery-harness

Key features

Run abnormal-order analysis
Generate compensation suggestions
Ingest and chunk in-memory documents
Search rules, documents, and cases
Execute evaluation cases synchronously
Record request traces and metrics

Use it for

Analyze abnormal delivery orders
Determine rule-based compensation suggestions
Test delivery workflows without a live model
Run evaluation suites on delivery scenarios

Alternatives to Delivery Harness

Harness Engineering — Improves agent output by shaping the environment with context and tools.
DeepSeek Harness — Builds on a plugin architecture to organize documents, analyze tables, and write code.
HarnessRouter — One API to manage agent runtimes, sandboxes, and execution lifecycles.
LongHorizon Harness — Maintains durable task state and executes reliable, verifiable workflows over many hours.
Exo Harness — Separates trusted state and security from agent logic for safe execution.
Harness2 — It recursively refines agent execution structures using parallel and sequential search.
Browser Harness — Agents edit their own helper scripts mid-task to improve task completion.
Fusion Harness — Run 2 to 5 models simultaneously against the same working directory with a single-writer invariant.
Sandbase Harness — It keeps agent infrastructure local, secure, and fully auditable.
Better Harness — Gathers project evidence and turns concrete gaps into prioritized findings.
Harness Meta-Skill — Generates specialized agent teams and skills for Claude Code from a single prompt.
Repository Harness — It prepares software repositories for safe coding agent integration.

All Delivery Harness alternatives