More Agent Autonomy Needs Stronger Guardrails: React 19 Linting on ESLint 10

A developer released @ternaus/eslint-plugin-react, an independent React 19 continuation for ESLint 10, after the upstream eslint-plugin-react failed to support the new ESLint version. The plugin narrows the rule set to React 19-specific checks that complement Biome, removing rules that duplicate Biome diagnostics or express style preferences. The developer's projects, including Albumentations.ai, sportscategory.info, and my-roots.me, use the plugin as part of stronger guardrails for coding agents.

The more autonomy I give coding agents, the more of a repository's expectations need to be executable. I do not want manual review to be the first place a predictable failure is discovered. My projects now have more pre-commit hooks, tests, deterministic checks for project conventions, and small reviewable commits. The checks turn expectations into pass/fail results an agent can act on; the commits keep failures narrow enough to diagnose. Linting is one of those guardrails. This story began when the React linting layer I relied on broke during an ESLint 10 upgrade. I maintain several React applications and wanted to upgrade them to ESLint 10. The upgrade stopped at eslint-plugin-react . ESLint 10 removed deprecated rule-context APIs https://eslint.org/blog/2026/02/eslint-v10.0.0-released/ . The current eslint-plugin-react@7.37.5 release declares support only through ESLint 9 and can crash under ESLint 10 with: TypeError: contextOrFilename.getFilename is not a function The upstream compatibility issue https://github.com/jsx-eslint/eslint-plugin-react/issues/3977 was opened on February 7, 2026. As of August 23, it is still open more than six months later. The pull request https://github.com/jsx-eslint/eslint-plugin-react/pull/4022 , opened on July 30, is also still open. That led me to release @ternaus/eslint-plugin-react https://github.com/ternaus/eslint-plugin-react : an independent React 19 continuation for ESLint 10. The package has a deliberately narrow support matrix: | Tool | Supported version | |---|---| | React | 19+ | | ESLint | 10 | | Biome | 2.5.8+ | | Node.js | 22.13, 24, or 26 | | ESLint config | Flat config | React 18, ESLint 9, and .eslintrc are outside the package contract. My projects use Biome as the primary formatter and linter. Biome handles general JavaScript, TypeScript, JSX, DOM, and most React checks. ESLint remains for checks that Biome does not provide, including framework plugins and several React 19 contracts. The setup came from real Next.js and React codebases behind Albumentations.ai https://albumentations.ai , sportscategory.info https://sportscategory.info , and my-roots.me https://my-roots.me : all presetMy projects need a smaller rule set: the React-specific checks that still add information after Biome has finished. I started from the upstream repository while preserving its Git history, MIT license, and attribution. The project is now maintained independently. At the commit where the fork began, the upstream plugin exported 104 rule modules. Its all preset enabled 102 active rules; the other two were deprecated. The recommended preset named 22 rules, although one of them, react/no-unsafe , was explicitly disabled, so it enforced 21. Carrying all of that forward would have preserved the package size without preserving a clear purpose. My target was narrower: React 19 checks that still add information after Biome has run. I reviewed the rules by the kind of decision they enforce: | Group | Decision | |---|---| | Same diagnostic already available in Biome | Remove from this plugin | | Formatting, naming, file layout, or team policy | Leave to Biome or the application | | Broad heuristics that need project-wide or type-aware evidence | Remove rather than report uncertain results | | React 18, classic config, parser workarounds, or obsolete React APIs | Remove from the React 19 contract | | React 19 correctness or a useful React-specific performance warning | Keep or implement | The first group accounted for exactly 20 rules, including jsx-key , no-danger , no-unknown-property , and self-closing-comp . The upstream rule support policy https://github.com/ternaus/eslint-plugin-react/blob/main/docs/upstream-rule-support.md records the complete one-to-one mapping to Biome. Rules such as jsx-sort-props , function-component-definition , and prefer-stateless-function were removed because they express style or team policy, not React 19 correctness. Rules such as no-unused-prop-types and no-unused-state were removed because a local AST heuristic cannot reliably answer a project-wide question. The remaining exclusions were legacy compatibility surface outside the package's stated support matrix. Four original rule IDs survived that review: jsx-no-constructed-context-values , no-deprecated , no-direct-mutation-state , and no-invalid-html-attribute . Among the upstream proposals relevant to this React 19 cutover, I evaluated three rule ideas that were still unmerged: undefined defaultProps on function components useState initializationI initially implemented all three. I then removed no-render-return-undefined : React 19 permits an undefined return, so forbidding it would be a team convention presented as a framework requirement. The other two became no-function-default-props and prefer-use-state-lazy-initialization . The first catches an API React 19 ignores; the second is a warning for avoidable render-time work. I also added or narrowed five rules around concrete React 19 behavior: controlled-form-requires-handler , jsx-no-key-after-spread , no-implicit-ref-callback-return , no-misspelled-lifecycle-methods , and no-prop-types . That left 11 rules in version 8.0.0. All 11 are in recommended : nine correctness rules are errors, while the two performance rules are warnings. A separate all preset would either duplicate recommended or differ only by severity, so the package does not expose one. By 8.0.0-rc.3 , the unit tests and package checks were green. Then I installed the plugin in Rooted and Albumentations.ai. That is where the useful failures started. The first class of failures was HTML attribute metadata. no-invalid-html-attribute rejected valid attributes such as alt , accept , name , loading , form , and value on <select , <option , and <textarea . The fixes arrived in three rounds: issue 21 https://github.com/ternaus/eslint-plugin-react/issues/21 and PR 22 https://github.com/ternaus/eslint-plugin-react/pull/22 , then issue 25 https://github.com/ternaus/eslint-plugin-react/issues/25 and PR 26 https://github.com/ternaus/eslint-plugin-react/pull/26 , and finally issue 29 https://github.com/ternaus/eslint-plugin-react/issues/29 and PR 31 https://github.com/ternaus/eslint-plugin-react/pull/31 . The final fix separated WHATWG HTML content attributes from React DOM properties instead of treating one metadata source as the whole contract. The second failure came from Next.js. eslint-config-next@16 builds flat config, but it still reads rules from the legacy-shaped react.configs.recommended.rules field. My package exposed only react.configs.flat.recommended , so configuration failed before ESLint could lint a file. Issue 24 https://github.com/ternaus/eslint-plugin-react/issues/24 and PR 27 https://github.com/ternaus/eslint-plugin-react/pull/27 added that read-compatible field without bringing .eslintrc support back into scope. I also documented the Yarn resolution needed because Next.js imports the plugin under the unscoped eslint-plugin-react name. Those integrations produced release candidates 4, 5, and 6. They also changed the test strategy: the final release checks the packed npm archive with publint , loads it from ESM, CommonJS, and TypeScript consumers, and exercises the supported Next.js configuration shape. CI covers Node.js 22.13, 24, and 26. The resulting package uses native ESM, ESLint flat config, and the familiar react/ rule namespace. The examples below use Yarn 4. Install Biome, ESLint 10, and the plugin: yarn add --dev @biomejs/biome@' =2.5.8' eslint@^10 @ternaus/eslint-plugin-react@^8.0.0 Enable Biome's full stable rule set and React domain: { "linter": { "domains": { "react": "all" }, "rules": { "preset": "all" } } } Then add the residual React rules to eslint.config.js : python import react from '@ternaus/eslint-plugin-react'; export default { files: ' / .{js,jsx,mjs,cjs,ts,tsx}' , ...react.configs.flat.recommended, }, ; If the glob includes TypeScript, configure a parser that supports TypeScript earlier in the flat config. Run both tools: yarn biome check . yarn eslint . Rule IDs remain under the react namespace: { rules: { 'react/no-deprecated': 'error', 'react/no-implicit-ref-callback-return': 'error', }, } eslint-config-next imports the package under the unscoped name eslint-plugin-react . With Yarn, map that dependency to the maintained package: { "devDependencies": { "@ternaus/eslint-plugin-react": "8.0.0" }, "resolutions": { "eslint-plugin-react": "npm:@ternaus/eslint-plugin-react@8.0.0" } } Keep both versions synchronized. This prevents eslint-config-next from installing the ESLint 9-only package alongside the ESLint 10 continuation. Because Next.js registers it under react , existing react/ rule IDs continue to work. This package does not cover every rule in the original plugin. If your configuration depends on rules such as react/jsx-sort-props , react/display-name , or react/prop-types , check the complete rule catalog https://github.com/ternaus/eslint-plugin-react/blob/main/docs/rules/README.md before migrating. The package also has no React Native-specific compatibility contract. DOM rules only analyze proven lowercase HTML elements.