browser-development-reference
Use when reviewing browser or web-platform work for known anti-patterns, weighing contested questions (Manifest V3, how much capability the web should have, the engine monoculture, Site Isolation's cost, JIT or no JIT, fingerprinting strategy, blocking ads by default, living standards vs versioned specs), checking whether an engine, standard, or privacy-policy claim is still current (snapshot verified August 2026), finding the primary sources, books, people, and channels, or needing the numbers, 'why is this page slow' triage order, and browser security review checklist. Companion to the other web-browser-development skills.
- License 许可
- MIT
- 最近更新
- 2026-11-15
- Stars
- ★ 5
- Forks
- 3
01 前言 Front Matter
- name
- browser-development-reference
- description
- Use when reviewing browser or web-platform work for known anti-patterns, weighing contested questions (Manifest V3, how much capability the web should have, the engine monoculture, Site Isolation's cost, JIT or no JIT, fingerprinting strategy, blocking ads by default, living standards vs versioned specs), checking whether an engine, standard, or privacy-policy claim is still current (snapshot verified August 2026), finding the primary sources, books, people, and channels, or needing the numbers, 'why is this page slow' triage order, and browser security review checklist. Companion to the other web-browser-development skills.
03 SKILL.md 正文
Browser Development: Anti-Patterns, Contested Questions, Currency, and Canon
Part 5 of 5 of the Web Browser Development reference (plugin
web-browser-development), covering §15–§20. Sibling skills:browser-engine-architecture-and-networking(§0–§3),browser-rendering-pipeline(§4–§7),browser-security-and-privacy(§8–§9),browser-extensions-platform-and-standards(§10–§14). Section numbers are shared across the set; a reference written as §N →skillpoints into that sibling skill.Currency: Verified August 2026. See §17 below for the currency snapshot and what goes stale first.
How to read this. Reference, not tutorial. Sections are independent. Three markers:
- [DURABLE] — architecture, algorithms, or a constraint every engine has independently arrived at. Does not expire.
- [ENGINE] — specific to Blink/Chromium, Gecko, WebKit, or Servo/Ladybird. Verify against that project's source or docs.
- [CONTESTED] — the vendors genuinely disagree, often because their business models differ. Both cases given, including the commercial motive where it's load-bearing.
⚠️ GOTCHA boxes mark the mistakes that produce security holes, jank, or compatibility breakage that can never be undone.
The two framings that organize everything below:
- A browser is a hostile-input execution environment that must never say no. It runs untrusted code from anyone, on documents that are frequently malformed, and it is not permitted to refuse or crash. Every architectural decision — the multi-process model, the error-tolerant parser, the sandbox — descends from that.
- The web's compatibility constraint is stronger than any other platform's. You cannot break existing sites. Not the well-written ones, not the abandoned ones. This is why the platform accretes and rarely subtracts, and why "just fix it properly" is almost never available.
§15. Anti-Patterns
| Anti-pattern | Why | Instead |
|---|---|---|
| Trusting an origin claimed by a renderer | Compromised renderer = full cross-site read | Re-derive and validate in the browser process (§8.4 → browser-security-and-privacy) |
| Giving a renderer direct file/network/device access | Removes the sandbox's entire point | Broker through validated IPC |
| Unvalidated sizes/indices/handles over IPC | Memory corruption in the privileged process | Validate every field; fuzz the surface |
| Custom HTML error recovery | Compatibility divergence from the whole web | Implement the spec algorithm exactly (§4.1 → browser-rendering-pipeline) |
| Left-to-right selector matching | Pathological performance | Right-to-left + bloom filters (§5.2 → browser-rendering-pipeline) |
| Recomputing all style on any mutation | Jank | Invalidation sets (§5.3 → browser-rendering-pipeline) |
| Mutable layout tree holding inputs and outputs | Unpredictable incremental layout | Immutable fragment tree (§6.2 → browser-rendering-pipeline) |
| Doing scroll or transform animation on the main thread | Jank whenever JS is busy | Property trees + compositor thread (§6.1 → browser-rendering-pipeline) |
will-change on everything |
GPU memory exhaustion | Layerize deliberately |
| Interleaving geometry reads and style writes | Forced synchronous layout, N× cost | Batch reads, then writes (§7.3 → browser-rendering-pipeline) |
| Sniffing content into an executable type | XSS vector | Honour nosniff; restrict sniffing |
| Shipping an XSS filter as a security boundary | Bypassable; introduces its own bugs; all were removed | CSP + Trusted Types (§8.3 → browser-security-and-privacy) |
| Unpartitioned storage or cache | Cross-site tracking and timing side channels | Partition everything by top-level site (§9.2 → browser-security-and-privacy) |
| High-resolution timers plus shared-process secrets | Spectre | Coarsen timers; isolate by site (§8.6 → browser-security-and-privacy) |
| Adding a capability API without an entropy review | Permanent fingerprinting surface | Review entropy and permission model together (§9.3 → browser-security-and-privacy, §11 → browser-extensions-platform-and-standards) |
| Shipping a web feature without an exit plan | You can never remove it | Origin trial + use counters first (§14 → browser-extensions-platform-and-standards) |
| Prefixed CSS properties | -webkit- became a de facto standard other engines had to implement |
Flags and origin trials, never prefixes |
| Reverse-engineering another engine's quirks instead of specifying them | Perpetuates the divergence | Spec it, add WPT, take it to Interop (§13 → browser-extensions-platform-and-standards) |
| Treating accessibility as a post-architecture concern | It crosses the process boundary; retrofitting is expensive | Design the a11y tree with the process model (§12 → browser-extensions-platform-and-standards) |
| Assuming the network service is trusted | It parses hostile input for all sites at once | Sandbox and memory-harden it (§3.1 → browser-engine-architecture-and-networking) |
§16. Contested Questions
16.1 Manifest V3. §10.2 → browser-extensions-platform-and-standards. Real security argument; real conflict of interest; Firefox
demonstrating that supporting both is technically possible.
16.2 Capability APIs — how much should the web be able to do? §11 → browser-extensions-platform-and-standards. Chrome's "the web
should match native" versus Apple's and Mozilla's "every API is attack surface and
fingerprinting entropy." Both coherent; the disagreement is about risk appetite and,
inescapably, about business models.
16.3 The engine monoculture. For consolidation: one excellent open-source engine beats five mediocre ones; interop problems vanish. Against: 81% share means one vendor's priorities become web policy, "works in Chrome" replaces "follows the spec," and there's no check on capability expansion. Note that the counterweight is weakening — Gecko is at ~3% and Mozilla's revenue depends largely on a Google search deal.
16.4 Site Isolation's cost. §2.2 → browser-engine-architecture-and-networking. Chromium's own docs concede the process-count ceiling,
especially on Android. The emerging answer is memory-safe languages buying back the
isolation that processes were paying for.
16.5 JIT or no JIT. §7.1 → browser-rendering-pipeline. Ladybird's no-JIT position and Apple's JIT entitlement
restrictions on iOS come from the same reasoning: JITs are enormous exploit surfaces.
The cost is performance, and nobody has shown a no-JIT engine that's competitive on
JS-heavy sites.
16.6 Fingerprinting: randomize or uniformize. §9.3 → browser-security-and-privacy.
16.7 Should the browser block ads by default? Brave does; others don't. It's a security control (malvertising), a privacy control, and an existential threat to the ad-funded web simultaneously.
16.8 Living standards vs. versioned specs. Living standards track reality and never go stale; they also mean "conformant" is not a fixed target, which is hard for anyone building an independent implementation — a point Ladybird and Servo feel directly.
§17. Currency Snapshot — verified August 2026
| Thing | Status as of Aug 2026 | Decay risk |
|---|---|---|
| Engine share | Blink ~81%, WebKit ~14%, Gecko ~3% (desktop+mobile+tablet, ~May 2026). Remaining ~1%: Goanna, Flow, Servo, Ladybird | Low |
| Ladybird | Independent C++ engine (LibWeb/LibJS), non-profit-backed, ~7–10 FT engineers, no code from other engines, no search-deal funding. Alpha targeted 2026 (Linux/macOS), beta 2027, stable 2028. Roadmap (not shipped): Rust style/layout, sandboxing, GPU isolation, Wasm GC | High |
| Servo | Rust, embeddable, parallel; under Linux Foundation Europe; components (Stylo, WebRender) shipped in Firefox. Positioned as a webview, not a browser | Medium |
| Third-party cookies | ⚠️ Not being deprecated in Chrome. 22 Jul 2024 first reversal; 22 Apr 2025 confirmed current approach with no standalone prompt. Safari/Firefox/Brave still block by default (~17–20% of traffic) | Medium |
| Privacy Sandbox | ⚠️ Mostly shut down. On 17 Oct 2025 Google retired most APIs (Topics, Protected Audience, Attribution Reporting, Private Aggregation, IP Protection, Related Website Sets) citing low adoption. Deprecated in Chrome 144 (Jan 2026), full removal targeted Chrome 150 (Jul 2026). Continuing focus: privacy-preserving measurement, FedCM, CHIPS | Medium |
| Manifest V2 | ⚠️ Essentially dead in Chrome. By June 2026 Chromium removed kExtensionManifestV2Disabled; Chrome 150/151 remove the last overrides. Edge and Opera follow. Full uBlock Origin unavailable on Chrome (uBO Lite only); available on Firefox and Brave. Firefox retains both MV2 and MV3 with blocking webRequest |
Medium |
| Interop | Interop 2026 is the fifth year (Apple, Google, Igalia, Microsoft, Mozilla). Interop 2025 finished with an overall score of 95 (from 25); Firefox went 46 → 99. 2026 areas: cross-document view transitions, attr(), contrast-color(), custom highlights, scroll-driven animations, scroll snap, shape(), Navigation API precommitHandler, scoped CustomElementRegistry, fetch streaming bodies, mobile testing |
Medium |
| Baseline | Trusted Types reached Baseline February 2026; also shape(), zstd, Navigation API. Newly available = all major engines; Widely available = +30 months |
Medium |
| Chrome memory safety | MiraclePtr expanding to Skia, ANGLE, Dawn, C++ iterators, std:: containers; MiracleObject on the GPU main thread targeting ~90% of UAFs there; "spanification" against OOB; UBSan -fsanitize=return default in release; PartitionAlloc in Skia; ChildProcessSecurityPolicy being migrated to Rust (Canary experiment) plus an initial Rust Mojo client; exploring an HTML/CSS/TypeScript top-level UI |
High |
| AI in browser security | Google's early-2026 Gemini-based agent harness found a sandbox escape that had survived >13 years in Chrome's codebase | High |
| Site Isolation ceiling | Chromium docs state plainly that sandboxing and site isolation are reaching their limits — processes are not cheap, especially on Android where extra processes cause background activities to be killed more often | Low |
Goes stale fastest: Chrome's memory-safety program and AI-assisted vuln finding;
Ladybird's alpha timeline; MV2 removal specifics; Privacy Sandbox removal milestones.
Essentially never stale: §2.1 → browser-engine-architecture-and-networking–2.4 (process model and IPC discipline), §4.1 → browser-rendering-pipeline (HTML
parsing), §5.1 → browser-rendering-pipeline–5.2 (cascade and selector matching), §6.1 → browser-rendering-pipeline–6.2 (pipeline and immutability),
§7.3 → browser-rendering-pipeline (event loop), §8.1 → browser-security-and-privacy–8.4 (security model), §15 (anti-patterns).
§18. The Canon
18.1 Primary sources — overwhelmingly the best material
- The specifications themselves, and they are unusually readable:
html.spec.whatwg.org (the parsing algorithm in §4.1 →
browser-rendering-pipelineis worth reading in full), dom.spec.whatwg.org, fetch.spec.whatwg.org, url.spec.whatwg.org, the CSS specs atdrafts.csswg.org, and HTML Standard §"event loops." - Chromium documentation —
chromium.organdchromium.googlesource.com/chromium/src/+/main/docs/. Specifically: the Site Isolation design doc,process_model_and_site_isolation.md, the memory safety page (unusually honest about limits), and the Chrome Security quarterly updates, which are the single best running account of browser security engineering anywhere. - RenderingNG series on
developer.chrome.com/docs/chromium/— overview, architecture, key data structures, BlinkNG, LayoutNG, VideoNG. Chris Harrelson et al. This is the best public description of a modern rendering engine's architecture, full stop. - Mozilla:
firefox-source-docs.mozilla.org, Mozilla Hacks (Stylo, WebRender, Quantum, and Fission write-ups), the Mozilla standards-positions repo. - WebKit blog (
webkit.org/blog) — especially the Interop and ITP posts. - web.dev and MDN — MDN is the de facto platform reference; web.dev carries Baseline and the Chrome rendering/performance material.
- Web Platform Tests (
web-platform-tests.org,wpt.fyi) and the Interop dashboard. - Ladybird (
ladybird.org, and the GitHub repo) and Servo (servo.org) — both are readable codebases in a way the big three are not, which makes them excellent learning material even if you never ship them.
18.2 Books and long-form
| Work | Why |
|---|---|
| Tali Garsiel & Paul Irish, "How Browsers Work" | The classic single-article overview. Dated in specifics, correct in shape |
| "Web Browser Engineering" (Panchekha & Harrelson) | Free online. Build a browser in Python, chapter by chapter. The best hands-on introduction that exists |
| Alan Grosskurth & Michael Godfrey, "A Reference Architecture for Web Browsers" | The academic framing of the component decomposition |
| Michal Zalewski, The Tangled Web | The best book on web security's actual model and its historical accidents |
| Ryan Barnett / OWASP materials; Google's Web Fundamentals security docs | Practical |
| Ilya Grigorik, High Performance Browser Networking | Free online. The networking reference (§3 → browser-engine-architecture-and-networking) |
| Lin Clark's cartoon deep-dives (Mozilla) | Stylo, WebRender, and Wasm explained better than anywhere else |
| "Inside look at modern web browser" (Mariko Kosaka, Chrome) | A clear four-part architecture series |
18.3 People and channels
Chris Harrelson (Blink rendering), Charlie Reis and Adrienne Porter Felt (Chrome security and Site Isolation), Andreas Kling (Ladybird — his development streams are unusually good teaching material), Lin Clark (Mozilla), Ilya Grigorik (networking), Alex Russell (platform, opinionated and worth reading precisely for that), Jen Simmons and Rachel Andrew (CSS layout), Anne van Kesteren (WHATWG specs), Michal Zalewski (web security). Follow the blink-dev and mozilla.dev.platform intent-to-ship threads if you want to see the platform being decided in public.
§19. Quick Reference
19.1 Numbers
- Frame budget: 16.6 ms @ 60 Hz, 8.3 ms @ 120 Hz — for everything.
- Engine share: Blink ~81%, WebKit ~14%, Gecko ~3%.
- Interop 2025 overall score: 25 → 95; Firefox 46 → 99.
- Baseline Widely available = Newly available + 30 months.
- Memory-safety bugs ≈ 70% of serious browser vulnerabilities historically.
- MiracleObject target: neutralize up to 90% of GPU-main-thread UAFs.
- ~17–20% of global traffic blocks third-party cookies by default regardless of Chrome.
19.2 "Why is this page slow?" — triage order
- Network: TTFB, blocking subresources, no preload, no compression, uncached.
- Main thread: long tasks, parse/compile cost, forced synchronous layout (§7.3 →
browser-rendering-pipeline). - Style: huge selector count, invalidation storms from class toggles on ancestors.
- Layout: layout thrashing, deep flex/grid nesting, table layout.
- Paint/raster: large paint areas, no layer promotion, or too many layers.
- Compositing: main-thread scroll (non-passive listeners), non-composited animations.
- GPU/memory: texture memory exhaustion, checkerboarding.
19.3 Browser security review checklist
- [ ] Does this feature let a renderer assert something the browser process trusts?
- [ ] Is every IPC field validated in the privileged process?
- [ ] Does it read cross-origin data into a renderer that shouldn't have it? (CORB/ORB)
- [ ] Does it add a timing signal usable for Spectre?
- [ ] Does it add fingerprinting entropy, and has that been quantified?
- [ ] Is it restricted to secure contexts? Permissions-Policy delegable?
- [ ] Is storage partitioned by top-level site?
- [ ] What is the permission UX, and can it be spoofed by page content?
- [ ] Does it parse untrusted binary input, and in what language, in which process?
- [ ] Is there an origin trial, a use counter, and a removal path?
§20. Sources and Method
Method. Narrative (not systematic) review. Durable material — §2 → browser-engine-architecture-and-networking (process model and IPC
discipline), §4.1 → browser-rendering-pipeline (HTML parsing), §5 → browser-rendering-pipeline (cascade and matching), §6 → browser-rendering-pipeline (pipeline architecture),
§7.3 → browser-rendering-pipeline (event loop), §8.1 → browser-security-and-privacy–8.4 (security model), §12 → browser-extensions-platform-and-standards (accessibility), §15 (anti-patterns) —
is synthesized from the specifications and from engine documentation listed in §18. Every
time-sensitive claim (engine shares, project status, policy changes, security-program
specifics) was verified against a primary or near-primary source in August 2026 and is
flagged in §17 with a decay-risk rating. Where vendors genuinely disagree, §16 presents
both cases — and names the commercial interest where it is material to the disagreement,
because in this domain it usually is.
Search log (August 2026): browser engine landscape and Ladybird/Servo status · Privacy Sandbox and third-party cookie policy · Interop 2026 and Baseline · Chrome site isolation, sandboxing, and memory safety · Manifest V3 and the extension platform · RenderingNG, BlinkNG, and LayoutNG architecture.
Primary and near-primary sources consulted (selected):
- chromium.org — Site Isolation design doc and overview;
process_model_and_site_isolation.md; the memory safety page; Chrome Security quarterly updates - developer.chrome.com — the RenderingNG series (overview, architecture, key data structures, BlinkNG, LayoutNG), Chris Harrelson et al.
- blog.google / Google Security — "Stronger with every update: How we're making Chrome and the web safer in the AI Era" (2026); SecurityWeek coverage of the 13-year-old flaw found by Google's agent harness
- privacysandbox.google.com — "Next steps for Privacy Sandbox and tracking protections in Chrome"; contemporaneous reporting on the Oct 2025 API retirements
- web.dev — Interop 2026 announcement, Baseline monthly digests; webkit.org — "Announcing Interop 2026"; hacks.mozilla.org — "Launching Interop 2026"; web-platform-tests/interop 2026 README
- ladybird.org and the Ladybird W3C/TPAC session description; servo.org and Servo's W3C presentation materials
- ublockorigin.com and the uBlock wiki on MV2 deprecation; Mozilla Blog on its MV3 approach; Neowin, The Register, PCWorld on the MV2 flag removals
- WHATWG and W3C specifications as cited throughout
Confidence statement. High confidence in §2–§8 → browser-engine-architecture-and-networking, browser-security-and-privacy and §11–§15 → browser-extensions-platform-and-standards, §18–§19 — these rest on
specifications and first-party engine documentation, much of it unusually candid.
High confidence in §17's verified items as of the stated date. Moderate confidence
in the engine market-share figures in §1.1 → browser-engine-architecture-and-networking and §19.1: browser share is measured by
analytics vendors with known sampling biases, varies by region and device class, and the
~81/14/3 split should be read as approximate and directional. Moderate confidence in
§9.1 → browser-security-and-privacy's timeline granularity — the third-party-cookie story reversed twice and secondary
sources disagree about which date was "the" reversal, so the dates here follow Google's own
posts and are given as a sequence rather than a single event. The Ladybird timeline in
§1.1 → browser-engine-architecture-and-networking and §17 is a project target, not a shipped fact, and such targets have slipped
before across this entire field.
04 同仓库更多 Skills
white-paper-writing
Use when writing white papers, technical reports, policy documents, thought leadership pieces, or any long-form authoritative document. Triggers on requests to write, structure, outline, or review white papers, technical reports, position papers, policy briefs, or evidence-driven arguments. Also use for B2B content marketing strategy, technical documentation requiring authority, academic-style reports, or any document that must educate while persuading. Use when someone asks "how should I structure this" for a long-form document.
prose-technical-mechanics
Use when editing or writing prose at a technical/mechanical level: choosing between Anglo-Saxon and Latinate diction, placing information in topic vs stress positions, writing periodic vs cumulative sentences, handling sentence rhythm, punctuation as logic, transitions, paragraph structure, or phonesthetics. Use when someone wants to understand WHY prose works or doesn't, not just surface-level style advice. Triggers on word choice, sentence structure, emphasis, rhythm, diction, or any deep writing mechanics question.
prose-mechanics
Use when writing, editing, or analyzing prose style; when trying to avoid generic AI-sounding writing; when applying specific prose styles such as journalistic, academic, narrative, KJV biblical, Hemingway plain, Latinate/Jamesian, Orwellian, Augustan, Victorian, Early Modern (Bacon, Browne, Milton), or free indirect discourse; or when coaching someone on writing voice and mechanics. Triggers on any request to improve, analyze, or write in a specific prose style, to fix "AI-sounding" writing, to make prose more vivid or concrete, to vary sentence rhythm, or to understand how a particular writer achieves their effects.
narrative-structure
Use when structuring stories, pitches, presentations, white papers, or any persuasive narrative; analyzing the hero's journey in fiction or non-fiction; applying StoryBrand for business communication; designing narrative arcs for thought leadership; or understanding Vogler, Snyder, Propp, or Campbell frameworks. Triggers on any narrative structure, storytelling, pitch structure, StoryBrand, hero's journey, three-act structure, Freytag, Save the Cat, narrative arc, or "how should I structure this story/pitch/presentation" request.
legal-brief-writing
Use when writing legal briefs, motions, or any court filing; when analyzing judicial cognition and persuasion; when applying elite brief-writing techniques from Roberts, Clement, Kagan, Garner; or when structuring legal arguments for maximum persuasion. Also use for autistic/neurodivergent advocates who need to understand how their natural communication style maps to elite legal advocacy. Triggers on any legal writing, brief drafting, motion preparation, or legal argument structuring request.
journal-publishability-criteria
Use whenever a research paper, article or preprint is being written, revised, reviewed or published, and specifically before every publication of docs/paper.md. Covers what legitimate journals actually gate on, in order of gating — the editor's desk screen (scope, novelty, package problems), what reviewers judge (methodological soundness, claims the evidence can carry, statistical and qualitative reporting), EQUATOR reporting guidelines, the ICMJE authorship criteria, COPE's view of text recycling, the near-universal generative-AI policy (no AI author, disclosure, author responsibility), TOP 2025 transparency standards, how to verify a journal is legitimate, submitting without an institution, and a pre-submission checklist. Triggers on requests to evaluate, review, prepare or submit a paper, to judge publishability, to write a declarations or data-availability section, to disclose AI use, or to check a journal.