Rust Web Frameworks 2026

Meta-prompt quality — ship (4.8/5)

dimension	score	reason
goal_fidelity	5	The meta-prompt strictly adheres to the GOAL of surveying open-source Rust web frameworks for maturity, async support, and ecosystem. It avoids drifting by explicitly excluding non-web-framework repositories and irrelevant data (e.g., Axum Wikipedia page). The constraints (open-source only) are respected.
capability_fit	4	The meta-prompt provides clear, essential instructions for data grounding (e.g., star counts, commit dates, scraped content) and evaluation criteria (maturity, async support, ecosystem). It avoids gratuitous filler by focusing on actionable data points. However, it could explicitly mention handling edge cases like archived repositories or forked frameworks, though this is minor.
reasoning_transparency	5	The meta-prompt justifies inclusions/exclusions (e.g., why `libunftp` is excluded) and success criteria (e.g., confidence scoring rubric). It explains how to assess each dimension (maturity, async support, ecosystem) using specific data points. The self-scoring section is particularly clear.
actionability	5	The instructions are highly actionable: they specify exact data sources to use, how to evaluate each framework, and the output schema. The no-op check passes, as executing this would produce a correct, grounded survey. The exclusion of non-frameworks and irrelevant data ensures correctness.
ambiguity_handling	5	The meta-prompt surfaces ambiguities (e.g., frameworks with no scraped content get lower confidence scores) rather than guessing. It explicitly addresses potential misinterpretations (e.g., Axum Wikipedia page) and provides rules for edge cases (e.g., Wasm frameworks' async support).

Output quality — ship (4.0/5)

dimension	score	reason
format_compliance	5	The OUTPUT JSON adheres closely to the required format, including all specified fields (name, repo, what_it_does, maturity, confidence, why_included) and additional fields like async_support and ecosystem. The schema is consistent across all items.
accuracy	3	The OUTPUT is largely accurate where scraped data is available (e.g., Dioxus, Yew, Actix Web, Axum). However, it includes inferences for frameworks without scraped data (e.g., Rocket, Warp, Tide), which are reasonable but not directly supported by the SOURCE DATA. Some claims (e.g., async support for lesser-known frameworks) lack grounding. No outright fabrications, but over-reliance on ecosystem knowledge risks inaccuracies.
self_score_calibration	4	Confidence scores are well-calibrated for frameworks with scraped data (0.8-0.9) and appropriately lower for frameworks without (0.1-0.4). The 'why_included' fields reflect the evidence strength. No uniform 0.99 scores; calibration is reasonable.
completeness	5	The OUTPUT includes 12 frameworks, covering all relevant repositories from the SOURCE DATA. All required fields are populated, even for frameworks with minimal data. No empty/near-empty results; completeness is high.
usefulness	4	The OUTPUT is highly actionable, providing detailed maturity assessments, async support, and ecosystem context for well-known frameworks. Even lesser-known frameworks are included with clear confidence caveats. The 'why_included' field adds transparency. Usefulness is strong despite some gaps in data.