pnpm Adds Native Workspace Releases, Oxlint Type-Aware Linting Goes Stable
pnpm's native workspace release support removes a common pain point for monorepo teams that previously relied on third-party tooling. Oxlint's stable type-aware mode means Rust-speed linting that actually understands your types is now production-ready, a direct performance upgrade over TypeScript's own compiler for lint workflows.
pnpm's latest batch of releases adds native workspace version publishing and a `doctor` command that diagnoses broken installs. Node.js launched a preview of its redesigned API docs site, now with search. Oxlint's type-aware linting — a long-running experimental feature — is now stable, bringing faster, Rust-based lint feedback that understands TypeScript types.
On the library side, TanStack released two tiny new packages: a 4.9 KB Markdown parser with serializable ASTs and an 8 KB syntax highlighter covering 25 languages. Both are already running in TanStack's own production. MobX 7.0 drops deprecated APIs, shrinks the bundle, and moves entirely to Proxy-based observables.
A reverse-engineering deep-dive into Claude Code v2.1.88 walks through the tool's entry architecture line by line, while a separate critique argues that Spec-Driven Development — writing a spec and handing it to a coding agent — breaks down for most real-world projects.
pnpm's doctor command signals that package manager tooling is maturing beyond basic installs into diagnostics, a space historically left to manual debugging.
Oxlint reaching stable type-aware linting closes a significant gap: developers could already lint quickly with Rust-based tools, but only now can they do so with full type information, which catches an entirely different class of bugs.
The critique of Spec-Driven Development highlights a tension that will only grow as coding agents improve — the gap between a spec's idealized description and the messy reality of codebases is not something better prompt engineering alone can close.
TanStack shipping its own Markdown parser and highlighter, and immediately using them in production, follows the same pattern the ecosystem saw with its router and table libraries: build the tool you need, open-source it, and let adoption follow.