Why We Stopped Optimizing Our Dev Stack and Started Removing Tools — The Anti-Stack That Cut Context Switching 40%

TL;DR: “Platform fatigue” in tools + “optimization fatigue” in wellness = same pattern. We ran a tool audit: 23 tools → 12. Killed 11, consolidated 4 into 2 Claude Code Routines. Context switches/day: 47 → 28. Tool audit spreadsheet →

The Context

Two-dev team, Basso Digital. June 2026: Cross-brand pattern spotted — Tool Crucible devs complaining about update dread, subscription fatigue; Well Worth It Lab audience rejecting supplement stacks, over-optimized routines. Unified insight: More isn’t better. Started monthly “Anti-Stack” review: what can we remove this month?

What We Tested

Category	Before (Tools)	After (Tools)	Removed	Consolidated
AI Coding	Copilot, Cursor, Windsurf, Codex, Cline, Continue	Copilot, Cursor, Claude Code, Ollama	Windsurf, Codex, Cline, Continue	—
Terminal	Warp, Ghostty, iTerm2, tmux, zellij	Ghostty + tmux	Warp, iTerm2, zellij	—
Database GUI	TablePlus, DBeaver, pgAdmin, Supabase Dashboard	DBeaver only	TablePlus, pgAdmin, Supabase Dashboard	—
API Testing	Postman, Insomnia, Hoppscotch, curl scripts	Bruno (local) + curl	Postman, Insomnia, Hoppscotch	—
Docs	Notion, Obsidian, GitBook, HackMD, Google Docs	Obsidian + Quartz	Notion, GitBook, HackMD, Google Docs	—
Routines	0 (manual prompts)	12 Claude Code Routines	—	4 workflows → 2 Routines

The Pivot Point

June 12, 2026: Counted context switches in a typical day — 47 (Cmd+Tab between 12 apps). Realization: “Optimization” had become procrastination. Each tool added cognitive load. Ran first Anti-Stack session: “If we didn’t have this tool, what would break?” 11 tools had no honest answer.

What We Use Now

Anti-Stack Protocol (monthly, 30 min):

# 1. Inventory (auto-generated)
./scripts/tool-inventory.sh > tool-inventory-$(date +%Y%m).md

# 2. For each tool, answer:
# - Last used? (auto from shell history)
# - Unique capability? (not covered by another tool)
# - Weekly time saved vs. maintenance cost?
# - Exit cost if removed today?

# 3. Decision matrix:
# KEEP: Unique capability + weekly value > maintenance
# CONSOLIDATE: Overlaps with another; merge into Routine/script
# KILL: No unique value; remove + document migration

Current Anti-Stack (12 tools, all justified):

1. VS Code — Universal shell (editor, terminal, git, extensions)
2. Ghostty + tmux — Terminal multiplexer; sessions persist
3. Copilot — Inline completions (speed)
4. Cursor — Multi-file refactors (Composer)
5. Claude Code — Architecture, verification, Routines (primary)
6. Ollama — Local LLM fallback (privacy)
7. DBeaver — Universal DB GUI (Postgres, MySQL, SQLite)
8. Bruno — Local API client (git-friendly collections)
9. Obsidian + Quartz — Docs/wiki (Markdown source, static publish)
10. Coolify — Self-hosted PaaS (Git push deploy)
11. PostHog (self-hosted) — Analytics (event + session replay)
12. Plane (self-hosted) — Issue tracking (projects, cycles, views)

When You’d Choose Differently

• Larger team (5+): Specialized tools pay off; shared context reduces switching cost
• Design-heavy work: Figma, v0.dev, Storybook justify their place
• Enterprise compliance: Some tools required for audit trails (SOC2, HIPAA)

Tool Crucible Rating

Dimension	Rating (1–5)	Notes
Overall	5	Subtraction > addition; the stack that doesn’t distract
Ease of Use	5	Fewer tools = less config, less updates, less breakage
Value	5	40% fewer context switches = 2h/day recovered
Support	5	Maintaining 12 tools is trivial vs 23

---

This is part of our Digital Minimalism / Tool Consolidation series. See audit template: Anti-Stack Tool Audit