●FORK — Claude Code 2.1.212 changes what /fork does: it copies your conversation into a new background session with its own row in claude agents, so you can keep working. The old in-session subagent is now /subtask●LIMITS — WebSearch calls are now capped at 200 per session by default, and subagent spawns get the same 200 ceiling, so a runaway search or delegation loop stops on its own●MCPBG — MCP tool calls running past two minutes now move to the background automatically, keeping the session usable. Tune the threshold with CLAUDE_CODE_MCP_AUTO_BACKGROUND_MS●PLANFIX — Fixed plan mode auto-running file-modifying Bash commands such as touch and rm without a permission prompt or an SDK canUseTool callback●SONNET5 — Claude Sonnet 5 is running on introductory pricing of $2 per million input tokens and $10 per million output. After August 31 it moves to $3 and $15●IPO — Bankers are reportedly lining up investor meetings for Anthropic ahead of a possible public listing as soon as October●FORK — Claude Code 2.1.212 changes what /fork does: it copies your conversation into a new background session with its own row in claude agents, so you can keep working. The old in-session subagent is now /subtask●LIMITS — WebSearch calls are now capped at 200 per session by default, and subagent spawns get the same 200 ceiling, so a runaway search or delegation loop stops on its own●MCPBG — MCP tool calls running past two minutes now move to the background automatically, keeping the session usable. Tune the threshold with CLAUDE_CODE_MCP_AUTO_BACKGROUND_MS●PLANFIX — Fixed plan mode auto-running file-modifying Bash commands such as touch and rm without a permission prompt or an SDK canUseTool callback●SONNET5 — Claude Sonnet 5 is running on introductory pricing of $2 per million input tokens and $10 per million output. After August 31 it moves to $3 and $15●IPO — Bankers are reportedly lining up investor meetings for Anthropic ahead of a possible public listing as soon as October
Claude Cowork's true power emerges when you combine skills (reusable workflows) with scheduling (time-based or event-driven execution). What starts as simple automation—"run this task daily"—scales to enterprise workflows: conditional logic, parallel processing, error recovery, and cross-team coordination. Here's how.
What follows includes implementation code, benchmarks, and practical content we hope you'll find useful. This site runs without ads — server and development costs are supported entirely by members like you. If it's been helpful, we'd be truly grateful for your support.
Goal: Real-time KPI tracking with instant alerts on anomalies.
skill: "KPI Health Monitor"schedule: cron: "0 */6 * * *" # Every 6 hoursworkflow: 1. Fetch: - Google Analytics: page views, conversion rate - Stripe: revenue, churn rate - Custom APIs: feature adoption 2. Compare vs. 30-day average 3. If deviation > 20%: → 🔴 Slack alert: "Revenue -35% vs. avg" → Suggest root causes (email issue? UI bug?) → Mention relevant team leads 4. Auto-log all metrics to Google Sheets 5. End of month: Auto-generate "KPI Report"
Output example:
🔴 ALERT: Conversion Rate Drop
Current: 2.1% | Average: 3.2%
Δ: -34%
Likely cause: Email delivery issue (Gmail bounce rate up)
→ @Support team notified on Slack
Use Case 4: Auto-Generated Meeting Minutes
Goal: Meeting ends → Minutes auto-generated in Notion with action items auto-assigned.
skill: "Meeting Minutes Auto-Gen"trigger: type: "webhook" source: "google_meet" event: "meeting_ended" on_trigger: "immediate"workflow: 1. Google Meet recording auto-saves 2. Audio transcribed (Deepgram) 3. Claude generates structure: - Attendees - Topics discussed - Decisions made - Action items (owner, deadline) 4. Auto-create Notion entry: "Meetings" DB 5. Extract action items 6. Slack: Notify each owner with their task + deadline
Sample output:
【Meeting: Q1 Strategy Review】
Date: 2026-03-27, 10:00 AM
Attendees: CEO, CTO, Marketing Head
【Decisions】
✓ Ship AI feature in Q2
✓ +20% marketing budget
✓ Deprecate legacy API
【Action Items】
□ CTO: Write AI design doc (Due: 4/5)
→ @CTO notified on Slack
□ Marketing: Budget allocation (Due: 3/30)
→ @Marketing notified on Slack
Use Case 5: Smart Customer Follow-Up (Behavioral)
Goal: Auto-identify customers needing follow-up based on behavioral signals.
skill: "Smart Customer Follow-Up"schedule: cron: "0 10 * * *" # Daily, 10 AMworkflow: 1. Fetch from Stripe + HubSpot: - Last purchase date - Last email open - Contract renewal date 2. Classify each customer: 【30 days no purchase】→ Reminder email 【90 days no purchase】→ Flag for sales team 【30 days to renewal】→ Renewal notice 【Churn risk score > 0.7】→ Escalate to CEO 3. Auto-generate "Follow-Up Needed" list (Google Sheets) 4. Slack #sales: "15 customers need follow-up today" 5. Link to Sheets for easy access
Chaining MCPs (Connectors) for Complex Workflows
Multi-Connector Flow Example
HubSpot ──→ Slack ──→ Google Sheets ──→ Gmail ──→ Notion
↓ ↓ ↓ ↓ ↓
Fetch Ask Aggregate Send Log
data team & save email results
Fix: Admin re-authenticates manually in Cowork UI.
Error Type 2: API Rate Limit / Timeout
error_handling: - type: "api_timeout" action: "retry" retry_count: 3 retry_delay: 300 # Wait 5 min, try again max_wait: 30000 # Don't wait more than 30 sec per attempt
Error Type 3: Timeout (No Response)
error_handling: - type: "wait_timeout" wait_seconds: 1800 # Wait 30 min for Slack response fallback: "skip_step" # Continue even if no response log: "Slack timeout - no sales team response"
Best Practices
1. Idempotency (safe to re-run)
# ❌ Bad: Appends every time (duplicates risk)action: "append_row"# ✅ Good: Updates existing or creates newaction: "upsert"match_on: "date" # If date exists, update; else create
Cowork's skills + scheduling framework handles enterprise-grade automation: sales reporting, standups, KPI monitoring, meeting minutes, customer follow-ups. All without code. All with natural-language configuration.
The key to production reliability: error handling, logging, and periodic human review. Automation is powerful—but "silent failures" are dangerous. Audit your workflows weekly.
Start small (one skill, one schedule). Validate. Then layer on complexity.
Share
Thank You for Reading
Claude Lab is ad-free, supported entirely by members like you. We publish practical guides daily with implementation code, benchmarks, and production-ready patterns. If you've found it useful, we'd love to have you on board.