# IsuruMaduranga/harness-engineering-101

> A book series explaining how to build AI agents from first principles, treating the LLM as a stateless brain and the harness as the program that manages the conversation array. It grows a ~300-line toy agent in plain Python across 16 chapters.

- Magnitude: 6.0 out of 10 — Early signal
- Stars: 173 total · +127 stars measured 2026-10-19, 12:58–14:32 UTC, ≈ 176 by evening
- Star trust: star growth looks organic
- Category: AI agents · Language: Python · Created: 2026-09-06 · Last push: 2026-10-19
- GitHub: https://github.com/IsuruMaduranga/harness-engineering-101 · Page: https://gitnova.dev/en/r/IsuruMaduranga/harness-engineering-101

## Useful for

- Read the chapters in order to understand how an agent loop works under the hood
- Run the toy harness from src/harness/ to experiment with tool calls and context budgets
- Build the book locally with mdbook to read it offline with rendered diagrams

## Why it’s here

- Star-counter measurements on 2026-10-19 (UTC), 12:58–14:32: 46 → 173 stars (+127). This is the change over that interval.
- Estimated end-of-day forecast: about +176 stars, using observed gains and the previous day.
- Before this, the repository barely got any stars — about 0 per day.
- The repository is 43 days old and already has 173 stars.
- Hacker News: “Harness Engineering 101: The LLM is the brain, the harness is the body” — 96 points, 2 h ago.

## Star trust

Star growth looks organic.

Star-trust labels are heuristics based on the repository’s behavior, not a check of every stargazer.

## Numbers

- Forks: 4
- Issues and pull requests: 0
- Watchers: 1
- Average over the last week: 25 per day
- Usual pace: 0 per day
- Stars in the last hour (measured): 54

## Stars per day, last 30 days (oldest → newest, today is partial)

2026-09-20 … 2026-10-19: 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 127

## Hacker News

- Harness Engineering 101: The LLM is the brain, the harness is the body — 96 points, 46 comments: https://news.ycombinator.com/item?id=50151390

## Spotted in now

- Spotted on Hacker News

## More in this category

1. **google-research/harness2** — 9.8 · Breakout · AI agents · Python · +3,860 stars measured 2026-10-19, 00:05–14:32 UTC, ≈ 6,237 by evening
   A framework for recursive optimization of AI agent harnesses (prompts, tools, memory, skills) without fine-tuning model weights, with JobBench, LAB and WorkBuddy-Bench benchmarks.
   Full card: https://gitnova.dev/en/r/google-research/harness2.md
2. **MadsLorentzen/ai-job-search** — 9.3 · Breakout · AI agents · Python · +2,685 stars measured 2026-10-19, 00:05–14:32 UTC, ≈ 4,216 by evening
   A job application framework built on Claude Code: evaluates postings, tailors CVs, and writes cover letters. Runs locally, focused on the Danish job market.
   Full card: https://gitnova.dev/en/r/MadsLorentzen/ai-job-search.md
3. **anomalyco/opencode** — 8.3 · Breakout · AI agents · TypeScript · +776 stars measured 2026-10-19, 00:05–14:32 UTC, ≈ 1,262 by evening
   Open source AI coding agent with terminal and desktop interfaces. Ships with a build agent (full edit and command access) and a read-only plan agent for code analysis.
   Full card: https://gitnova.dev/en/r/anomalyco/opencode.md
4. **alchaincyf/huashu-motion** — 7.6 · Early signal · AI agents · JavaScript · +544 stars measured 2026-10-19, 00:05–14:32 UTC, ≈ 838 by evening
   A skill for coding agents (Claude Code, Codex) that turns a product description into code-driven animated launch films, composes music, and opens a browser editor for frame-level edits.
   Full card: https://gitnova.dev/en/r/alchaincyf/huashu-motion.md
5. **finereli/whatsapp-mcp** — 7.2 · Early signal · AI agents · TypeScript · +229 stars measured 2026-10-19, 00:13–14:32 UTC, ≈ 336 by evening
   A read-only MCP server on Cloudflare Workers that lets your AI agent search and read your own WhatsApp messages. Your phone stays the source of truth; the server acts as a linked device and cannot send anything.
   Full card: https://gitnova.dev/en/r/finereli/whatsapp-mcp.md

---

Magnitude (0–10) measures how fast and how unusually interest in a repository is growing right now. It is not a quality score.
Days are UTC. “So far today” is a fact; “expected by the end of the day” is a forecast.
Summaries and use cases are written by an LLM (DeepSeek V4.1 Flash) from the README and may be inaccurate: verify specific claims (benchmarks, speed, hardware) in the repository itself.
Data as of 2026-10-19 14:47 UTC, updated every 30 minutes.