Harness Engineering 101
Comments on “Harness Engineering 101”
Claim-checked. Scored by all three judges, and its checkable claims were compared with the sources it cites. How far stories are checked
Isuru Wijesiri, who built WSO2's MI and BI Copilots and the Claude Code rebuild One Code, has published "Harness Engineering 101," a book arguing that agents are a loop around a stateless LLM API, with each feature patching a concrete failure. It covers tools, caching, context limits, subagents, MCP and RAG, and builds a roughly 300-line Python coding agent with no dependencies. The framing is the author's own, drawn from his experience, and is not independently sourced.
Why it ranked
No. 14 of 111 with a score of 0.67. The panel rated it highest on practical relevance (0.83) and new information (0.77), lowest on significance (0.56). Placed No. 14, No. 19 by score alone: the fourth AI & machine learning story of the day (−0.020). The placement terms move stories; they are not part of the score.
- AI judgeJudge 2claude-haiku-5-5
- A practical, well-structured tutorial on building AI agent harnesses from a simple loop; highly relevant to builders, though it is an educational book rather than news and its claims are experiential.
- AI judgeJudge 3gpt-5.6-luna
- A practical, code-oriented explanation of agent harness design, grounded in firsthand projects and an incremental toy implementation. It is broadly useful to AI and developer-tool builders, though the framing is partly pedagogical and some claims are presented as generalizations.
Each judge wrote its line after answering the rubric, with the site and the author still hidden. Every answer behind this is in the table under Score.
Score
0.67total under rubric v2.3
- Significance0.56
- Reader interest0.76
- Audience breadth0.66
- Practical relevance0.83
- Substantiated0.69
- Technical depth0.68
- New information0.77
- Counts against: Promotional0.09
- Counts against: Overclaim0.20
Each judge’s answer per question (0–1), their mean and spread, and what the question adds to the score (weight × mean).
| Question | Judge 1 | Judge 2 | Judge 3 | Mean | Spread | Weight | Contribution (weight times mean) |
|---|---|---|---|---|---|---|---|
| SignificanceHow much does this change what people in tech know or can do? | 0.60 | 0.45 | 0.62 | 0.56 | 0.17 | +0.24 | +0.134 |
| Reader interestWould a well-informed senior engineer or tech founder want to read this today, and likely pass it on? | 0.73 | 0.70 | 0.86 | 0.76 | 0.16 | +0.26 | +0.198 |
| Audience breadthHow many people who build technology does it concern? | 0.67 | 0.60 | 0.72 | 0.66 | 0.12 | +0.13 | +0.086 |
| Practical relevanceHow useful is it to someone who builds software or tech products? | 0.92 | 0.75 | 0.82 | 0.83 | 0.17 | +0.22 | +0.182 |
| SubstantiatedHow well is it backed? Consider whether it is the primary source, and whether its main claims come with data, code, a demo, a reproducible method or citations. | 0.79 | 0.50 | 0.78 | 0.69 | 0.29 | +0.05 | +0.034 |
| Technical depthHow technically substantive is it? | 0.73 | 0.60 | 0.70 | 0.68 | 0.14 | +0.05 | +0.034 |
| New informationDoes it contain information absent from the stories covered in the last 7 days (list provided)? | 0.90 | 0.60 | 0.82 | 0.77 | 0.30 | +0.05 | +0.039 |
| PromotionalIs it mainly marketing with little substance (a press release, ad or sponsored post with no real information)? A launch post that carries real news is not promotional. | 0.10 | 0.10 | 0.08 | 0.09 | 0.02 | −0.19 | −0.018 |
| OverclaimDo the headline or conclusions claim more than the body supports? | 0.23 | 0.20 | 0.18 | 0.20 | 0.05 | −0.11 | −0.022 |
| Sum of contributions | +0.668 | ||||||
| Claim-check penalty | 0.000 | ||||||
| Total, as published | 0.668 | ||||||
Spread is the highest minus the lowest judge answer. marks a spread above 0.40: the judges disagree, and the story is marked contested.
The badge for this ranking, for the author's page or anyone's: one line of HTML, and it links back here.
<a href="https://rubric.hrih.si/editions/2026-10-31/14/"><img src="https://rubric.hrih.si/badge/2026-10-31/14.svg" alt="Ranked #14 on Rubric, 31 Oct 2026" width="245" height="28"></a>Claim check
Claim check: 2 matches source · 4 no source given · 2 couldn’t check. Each claim is compared with the source the article itself cites; a flag is published only when both checkers agree.
No source givenGeneralization
An LLM API is stateless: every turn the full conversation is sent as a JSON array and text is returned.
The article gives no link or citation for it.
The article says “An LLM API is stateless. Every turn, you send the entire conversation as a JSON array and get text back”
The source says No source to compare with.
No source givenGeneralization
Everything the field calls agents is a set of patches to the one harness loop, each added because something concrete broke.
The article gives no link or citation for it.
The article says “Everything the field calls “agents” is a set of patches to that one loop, and each patch exists because something concrete broke”
The source says No source to compare with.
No source givenDate
The author started WSO2 MI Copilot, a coding agent for software integration, in October 2023.
The article gives no link or citation for it.
The article says “In October 2023 I started WSO2 MI Copilot, a coding agent for software integration”
The source says No source to compare with.
No source givenDate
WSO2 MI Copilot was shipped at WSO2Con 2024 and ran on GPT-3.5, GPT-3.5 Turbo, and later GPT-4.
The article gives no link or citation for it.
The article says “I led the work until we shipped it at WSO2Con 2024. It ran on GPT-3.5 and GPT-3.5 Turbo, and later GPT-4”
The source says No source to compare with.
Matches sourceVersion
One Code is a from-scratch rebuild of Claude Code on the open-source pi harness.
The cited source says the same thing.
The article says “I also built One Code, a from-scratch rebuild of Claude Code on the open-source pi harness”
The source says “One Code is an independent, open-source implementation built on [pi](https://github.com/earendil-works/pi). It recreates the coding workflow; features that depend on Claude Code's hosted or desktop services are outside its scope”
Source cited by the article: github.com/IsuruMaduranga/one-code
Couldn’t checkVersion
WSO2 BI Copilot does the same job as MI Copilot in the Ballerina programming language.
The source was unreachable, paywalled or too long, or the two checkers disagreed.
The article says “WSO2 BI Copilot, which does the same job in the Ballerina programming language”
The source says No passage could be compared.
Source cited by the article: wso2.com/integration-platform/docs/saas/editor/copilot
Matches sourceNumber
The toy harness grows into a working mini coding agent of about 300 lines by the epilogue.
The cited source says the same thing.
The article says “By the epilogue you’ll have a working mini coding agent in about 300 lines”
The source says “This series develops that one idea across 16 chapters, growing a ~300-line toy agent in plain Python as it goes”
Source cited by the article: github.com/IsuruMaduranga/harness-engineering-101
Couldn’t checkVersion
The toy harness uses Python 3.10+ with zero dependencies and calls the Anthropic Messages API by default.
The source was unreachable, paywalled or too long, or the two checkers disagreed.
The article says “It uses Python 3.10+ with zero dependencies (raw urllib) and calls the Anthropic Messages API by default”
The source says No passage could be compared.
Source cited by the article: github.com/IsuruMaduranga/harness-engineering-101/tree/main/src/harness
Also covered by
No other coverage was clustered with this story.
Elsewhere
Elsewhere: HN 1 pts, not on front page
Recorded for context only, checked 31 Oct 2026, 05:46 UTC. It played no part in the ranking.