CClaude Code Catalog
All Skills
Update Briefing

Claude Code Card News: Top 9

A curated card-news digest of recent official Claude and Claude Code updates with high practical impact.

#1Claude Code DocsResearch Preview

Claude Code Channels — Control Sessions via Telegram & Discord

Channels is a research preview feature that lets you push external events into a running Claude Code session. Connect a Telegram or Discord bot to your local session, and messages from your phone arrive as channel events — Claude acts on them and replies back through the same chat. Unlike standard MCP servers where Claude pulls data on demand, Channels actively push events into the session. Beyond chat bridging, Channels work as webhook receivers, routing CI results, error tracker alerts, and deployment pipeline events directly into a debugging session that already has your files open. Requires Claude Code v2.1.80+ and claude.ai login. Enterprise plans must explicitly enable Channels in admin settings.

Why It Matters

Channels formalizes the 'supervise from your phone while Claude works in the background' workflow. Where Remote Control means 'you drive the session from mobile,' Channels means 'external systems send Claude notifications.' Particularly powerful for DevOps automation where CI pipelines and monitoring alerts are received and acted on directly by Claude. Currently in research preview — requires Bun, opt-in per session with the --channels flag.

Read Official Source
#2AnthropicNew Feature

Claude Code Skills System Official Launch

The biggest structural change to Claude Code — the Skills system — is officially live. Define skills in SKILL.md files with YAML frontmatter, and get five bundled skills out of the box: /simplify (3-way parallel code review), /batch (worktree-parallel large-scale changes), /debug (session log analysis), /loop (cron scheduling), and /claude-api (instant API docs access). Advanced patterns include $ARGUMENTS parameter substitution, !`command` dynamic runtime context injection, and context: fork for isolated sub-agent execution. The agentskills.io open standard ensures cross-platform compatibility with Cursor, Windsurf, and other agents.

Why It Matters

The Skills system isn't just a feature addition — it's a paradigm shift in how Claude Code is used. Define a workflow once in SKILL.md, and the entire team reuses it with a single /command. The four-tier deployment hierarchy — personal (.claude/skills/), project (skills/), global (~/.claude/skills/), and environment ($CLAUDE_SKILL_DIR) — scales from small teams to enterprise organizations.

Read Official Source
#3AnthropicNew Feature

Auto Mode Research Preview Announced

Auto Mode lets Claude automatically evaluate and approve permission prompts for file edits and command execution, eliminating the need for manual approval on every action. Previously, every file write and shell command required user confirmation. In Auto Mode, Claude assesses each action's risk level and auto-approves safe operations. This eliminates dozens of approval clicks during long autonomous sessions, dramatically improving productivity for large-scale refactoring and migration tasks.

Why It Matters

Removes the biggest friction point in extended coding sessions — repetitive permission approvals. Currently a research preview for Enterprise plans, with Pro/Team expansion expected at GA. Combined with /batch, this opens the door to fully automated multi-hundred-file changes.

Read Official Source
#4Claude Code GitHubNew Feature

/loop Command + Cron Scheduling Introduced

The /loop command enables repeating prompts with flexible scheduling options: --count for fixed repetitions, --cron for cron expression-based periodic execution, and --duration with --interval for time-based loops. Each iteration automatically detects changes from previous runs and generates cumulative reports. Configurable stop conditions (on error, success, or change detection) are also supported. Voice mode gains pushToTalk keybinding for more convenient hands-free voice input.

Why It Matters

Marks Claude Code's evolution from a 'run and done' tool to a continuous monitoring system. Deploy health monitoring, daily code quality checks, and periodic data collection can now all run within Claude Code sessions — bridging the gap between development and operations.

Read Official Source
#5Claude Code GitHubNew Feature

/claude-api Built-in Skill + Voice 20 Language Support

The /claude-api built-in skill provides instant access to Claude API and Anthropic SDK documentation within your session. Reference API call patterns, tool_use implementation, and streaming handling while writing code. The $CLAUDE_SKILL_DIR environment variable enables centralized team skill management without copying files into each project. Voice STT adds 10 new languages (Russian, Polish, Turkish, and more) for a total of 20 supported languages.

Why It Matters

Completes the skills extensibility foundation. $CLAUDE_SKILL_DIR enables team-wide skill distribution without per-project duplication, /claude-api dramatically accelerates AI application development, and expanded Voice language support improves accessibility for non-English-speaking developers.

Read Official Source
#6AnthropicNew Service

Cowork & Plugins for Enterprise Launch

Claude Cowork embeds directly into existing work tools — Excel, PowerPoint, Slack, and Google Drive. Claude analyzes data within spreadsheets, generates slides, and drafts emails without app switching. Domain-specific enterprise plugins for finance (FactSet integration), HR (recruitment and onboarding automation), and engineering (code review and deployment management) launch simultaneously, enabling organizations to deploy Claude into existing workflows immediately.

Why It Matters

A turning point where Claude expands from 'developer tool' to 'enterprise-wide AI coworker.' Marketing, finance, and HR teams can now use Claude within their existing tools (Excel, Slack) without developer involvement. Combined with the $200M Snowflake partnership, Anthropic's enterprise market push is now fully underway.

Read Official Source
#7Claude Code GitHubNew Feature

Auto-memory Persistence Introduced

Claude Code automatically saves learned coding styles, project patterns, and user preferences to ~/.claude/projects/ and auto-loads them in subsequent sessions. Previously, users had to manually write rules in CLAUDE.md — now Claude remembers patterns it discovers while working. The /copy command is also added, enabling one-click clipboard copy of Claude's last response.

Why It Matters

If CLAUDE.md is 'explicit rules,' Auto-memory is 'implicit learning.' Claude understands your project more deeply with each session, reducing the need for repetitive instructions. Memory files are separated by project, making this especially valuable for developers juggling multiple codebases.

Read Official Source
#8AnthropicNew Feature

Remote Control — Continue Sessions from Mobile

Remote Control lets you continue a terminal Claude Code session from your phone or tablet, now available as a research preview. Native streaming eliminates the need for port forwarding or VPN setup. Sessions persist in the background and auto-reconnect when the host comes back online. Start with claude --remote-control to generate a QR code or URL for instant access from any device.

Why It Matters

Run long refactoring or deployment tasks and monitor them from a meeting room or coffee shop. Currently available for Max tier with Pro expansion planned. Combined with /loop, you can set up automated monitoring that alerts you on mobile when issues arise — enabling intervention without returning to your desk.

Read Official Source
#9AnthropicNew Service

Cowork Enterprise Plugins + Vercept Acquisition

Anthropic announces a major enterprise push for Cowork. Plugins connecting DocuSign (contract automation), FactSet (financial data analysis), and Google Drive/Gmail (document and email automation) are unveiled. Simultaneously, Anthropic acquires Vercept, a startup specializing in perception and interaction, to further improve Claude's computer use accuracy (94% on Pace Insurance Benchmark) for UI automation.

Why It Matters

Signals Anthropic's full expansion from developer tools into an enterprise agent platform. The Anthropic Marketplace with third-party apps from Harvey (legal) and Replit (development) is also taking shape, while the Vercept acquisition strengthens computer use capabilities for UI automation.

Read Official Source
#10AnthropicSecurity

Claude Code Security Announcement

A security-focused announcement officially documenting defensive coding workflows in Claude Code. Covers source code vulnerability scanning, dependency security auditing, automated security code review, and incident response patching. Includes concrete examples of running OWASP Top 10 checks, secret detection, and privilege escalation pattern detection within Claude Code sessions.

Why It Matters

Officially establishes Claude Code as an integrated security tool — no separate scanners needed. Combine with /simplify's code-quality agent for automated pre-merge security reviews, or integrate into CI/CD for continuous security monitoring.

Read Official Source
#11AnthropicModel Update

Claude Opus 4.6 Released — 1M Context Window

Anthropic's first major model release of 2026. The 1 million token context window can process entire corporate document libraries in a single session, with significant improvements in agentic coding, multi-step planning, and financial analysis. Agent Teams also launches alongside, officially supporting multi-agent patterns where multiple agents collaborate on large-scale tasks.

Why It Matters

The 1M context window is a game changer for analyzing large legacy codebases and full document reviews. Combined with Agent Teams, you can orchestrate multiple agents analyzing tens of thousands of lines of code, significantly expanding Claude Code's practical reach.

Read Official Source
#12AnthropicModel Update

Claude Sonnet 4.6 Launch — Default for Free/Pro Plans

The most capable Sonnet model ever is now the default for Free and Pro plans. Balanced improvements across coding, agent planning, and computer use, achieving 94% accuracy on the Pace Insurance Benchmark for human-level spreadsheet navigation and multi-step web forms. Available across both Claude Code and Cowork.

Why It Matters

Sonnet 4.6 offers an excellent performance-to-cost ratio for everyday development tasks. A 'two-tier model strategy' is recommended: Opus 4.6 for complex architectural decisions and large refactoring, Sonnet 4.6 for general coding, debugging, and reviews. Claude Code's Fast Mode also uses the same Opus model with faster output.

Read Official Source

Why This Claude Code News Digest Matters

Claude Code evolves quickly, and model or CLI updates can immediately affect team productivity. This page compresses official announcements and release notes into actionable highlights.

Checking this digest regularly helps keep prompts, CLAUDE.md policies, and automation pipelines current. Stability and auth/session changes are especially useful for reducing real-world operational issues.