Claude Code Daily Briefing - 2026-05-18

Release Summary

VersionDateKey Changes
v2.1.1435/15Plugin dependency management, marketplace token cost display, worktree.bgIsolation: "none", PowerShell Bypass

(No new release as of 5/18 — latest version is v2.1.143 from 5/15.)

Full release notes

New Features & Practical Usage

Claude Code Weekly Limits Up 50% — Through July 13 (announced 5/13)

Anthropic has increased Claude Code weekly usage limits by 50%. The boost applies to all Pro, Max, Team, and seat-based Enterprise users immediately, running through July 13, 2026 at 6PM PDT.

Key details:

The SpaceX Colossus 1 partnership (300MW+, 220,000+ NVIDIA GPUs) underpins this capacity expansion. Post-July 13 limits have not been announced.

ClaudeDevs announcement | Apidog analysis

Claude for Legal — 12 Practice-Area Plugins, 20+ Connectors (5/12)

Anthropic significantly expanded Claude for Legal. Claude Cowork now integrates with 12 practice-area specific plugins and 20+ legal platform connectors for specialized legal AI workflows.

12 plugin areas: Corporate, employment, regulatory, privacy, product liability, AI governance, intellectual property, litigation, and more.

20+ connectors: Thomson Reuters/CoCounsel, LexisNexis, iManage, NetDocuments, DocuSign, Ironclad, Box, Everlaw, Relativity, Consilio, and others.

Adoption: Freshfields, Quinn Emanuel, Holland & Knight are using Claude on live matters. Over 20,000 legal professionals signed up for Anthropic’s most recent legal webinar. Legal professionals have become the most engaged user group on Claude Cowork.

TechCrunch | Bloomberg Law

Developer Workflow Tips

The 60% Context Threshold Rule — More Effective Than /compact

Multiple practitioners have independently arrived at the same rule: reset your session when context exceeds 60%. It’s more effective than /compact because you control exactly what survives.

Method:

  1. Dump current plan and progress to a markdown file
  2. Run /clear to reset the session
  3. Start fresh with Claude reading that file
# Save progress then reset
# Ask Claude: "Write current plan and progress to progress.md"
# Then /clear
# New session: "Read progress.md and continue from where we left off"

/compact lets Claude auto-summarize, which can lose critical context. Manual reset ensures exactly the right information carries over.

Builder.io

The lessons.md Pattern — Let Claude Improve Itself

Every time Claude makes a mistake, after correcting it, have it add a rule to lessons.md. Over time, Claude literally teaches itself to be better at your specific project. Keep it separate from CLAUDE.md to avoid cluttering core configuration.

# lessons.md (example)
- Import paths in this project always use the @/ alias
- Test files use *.test.ts not *.spec.ts
- API responses always follow the { data, error } structure

Reference @lessons.md from CLAUDE.md so it auto-loads each session.

Claude Code Best Practices

Ecosystem & Plugins

Code with Claude London (5/19) D-1 + Extended (5/20) D-2

The London event is tomorrow. In-person registration is closed; livestream registration is open and recordings will be available afterward. The May 20 Extended session targets indie developers and early-stage founders with hands-on workshops and 1:1 office hours. Tokyo (6/10) and Tokyo Extended (6/11) registration is also open.

Code with Claude London | London Extended

/ultrareview — Cloud Code Review from CI/Scripts (Research Preview)

The claude ultrareview subcommand lets you trigger ultrareview from CI or scripts without an interactive session. Multiple reviewer agents analyze changes in parallel in a cloud sandbox, independently reproduce and verify each finding, and return results focused on real bugs rather than style suggestions.

# Run ultrareview from CI
claude ultrareview

Pro/Max get 3 free runs, then ~$5–$20 per run. Requires v2.1.86+.

Claude Code Docs

Community News

Minor Changes Worth Knowing

Interesting Projects & Tools

By subscribing, you agree to our Privacy Policy.