AI evals for agents: what they are and how to write them
Comments on “AI evals for agents: what they are and how to write them”
3 judges. Scored by all three judges. Only the day’s top stories get the claim check. How far stories are checked
This Evalshift guide explains what AI evals are and how agent evals differ from LLM evals, since agents must be graded on tool calls, arguments and resulting state rather than reply text alone.
Why it ranked
No. 64 of 107 with a score of 0.59. The panel rated it highest on practical relevance (0.87) and technical depth (0.81), lowest on significance (0.52). The judges disagreed most on new information (spread 0.58), so the story is marked contested. Placed No. 64, No. 55 by score alone: the 21st AI & machine learning story of the day (−0.080). The placement terms move stories; they are not part of the score.
- AI judgeJudge 2claude-haiku-5-5
- Practical tutorial on writing AI agent evals, actionable for builders, with a substantial vendor promotional section at the end and some overlap with known eval concepts.
- AI judgeJudge 3gpt-5.6-luna
- A practical, technically detailed guide to evaluating agent traces, outcomes, graders, flakiness, and CI regressions. It includes useful examples and metrics, though some claims are generalized and the later product section adds mild promotion.
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
0.59total under rubric v2.3Contestedjudges differ by 0.58 on new information, 0.53 on promotional
- Significance0.52
- Reader interest0.68
- Audience breadth0.62
- Practical relevance0.87
- Substantiated0.60
- Technical depth0.81
- New information0.63
- Counts against: Promotional0.32
- Counts against: Overclaim0.19
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? | 0.64 | 0.30 | 0.62 | 0.52 | 0.34 | +0.24 | +0.125 |
| Reader interestWould a well-informed senior engineer or tech founder want to read this today, and likely pass it on? | 0.74 | 0.45 | 0.84 | 0.68 | 0.39 | +0.26 | +0.176 |
| Audience breadthHow many people who build technology does it concern? | 0.67 | 0.50 | 0.70 | 0.62 | 0.20 | +0.13 | +0.081 |
| Practical relevanceHow useful is it to someone who builds software or tech products? | 0.98 | 0.75 | 0.88 | 0.87 | 0.23 | +0.22 | +0.191 |
| 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. | 0.64 | 0.50 | 0.67 | 0.60 | 0.17 | +0.05 | +0.030 |
| Technical depthHow technically substantive is it? | 0.98 | 0.70 | 0.76 | 0.81 | 0.28 | +0.05 | +0.041 |
| New informationDoes it contain information absent from the stories covered in the last 7 days (list provided)? | 0.88 | 0.30 | 0.72 | 0.63 | 0.58 , above the contested threshold | +0.05 | +0.032 |
| 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. | 0.07 | 0.60 | 0.28 | 0.32 | 0.53 , above the contested threshold | −0.19 | −0.060 |
| OverclaimDo the headline or conclusions claim more than the body supports? | 0.11 | 0.25 | 0.22 | 0.19 | 0.14 | −0.11 | −0.021 |
| Sum of contributions | +0.595 | ||||||
| Claim-check penalty | 0.000 | ||||||
| Total, as published | 0.595 | ||||||
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/2026-11-03/64/"><img src="https://rubric.hrih.si/badge/2026-11-03/64.svg" alt="Ranked #64 on Rubric, 3 Nov 2026" width="238" height="28"></a>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 3 Nov 2026, 05:47 UTC. It played no part in the ranking.