Sponsored by *Telnyx* (paid promotion) — Check it out here: https://telnyx.com/sign-up?utm_source=sponsored&utm_medium=youtube&utm_campaign=CamelCase
Code *CAMELCASE25* gets you $25 in free credits, no card required.

Rising AI bills and usage limits you hit way too early: I tested which token-saving tips for #ClaudeCode and #GitHubCopilot actually work — and which are more of a myth. Prompt caching, #MCP vs. CLI, #LSP vs. grep, /compact & co, model selection, and the famous #Caveman skill.

🍿 Chapters:
00:00 Intro
01:46 Prompt Caching
04:13 MCPs
09:18 LSP vs. grep
12:02 Compact & Co
13:27 Model Selection
14:55 Caveman

🔗 Links:
MCP token cost stats: https://policylayer.com/token-cost
Serena: https://github.com/oraios/serena
MCP Language Server: https://github.com/isaacphi/mcp-language-server
Caveman skill: https://github.com/JuliusBrussee/caveman

🧑‍💻 Further information:
Prompt caching (Anthropic docs): https://platform.claude.com/docs/en/build-with-claude/prompt-caching
How prompt caching works in Claude Code: https://www.claudecodecamp.com/p/how-prompt-caching-actually-works-in-claude-code
What if you don't need MCP?: https://mariozechner.at/posts/2025-11-02-what-if-you-dont-need-mcp/
MCP vs. CLI: https://www.scalekit.com/blog/mcp-vs-cli-use
The hidden cost of tool overload: https://deploystack.io/blog/mcp-token-limits-the-hidden-cost-of-tool-overload
LSP vs. grep token study: https://dev.to/daynablackwell/we-measured-it-lsp-saves-ai-agents-5-34x-tokens-vs-grep-427
JetBrains on the Caveman prompt: https://blog.jetbrains.com/ai/2026/07/speak-to-ai-agents-like-cavemen-tosave-tokens/
Caveman benchmark (a 6-line version wins): https://dev.to/jakguzik/i-benchmarked-the-viral-caveman-prompt-to-save-llm-tokens-then-my-6-line-version-beat-it-2o81

📜 Legal:
Music License Codes: VOINKK4VUN4TLJ6C, VX5YXVKM8I8FSXA8, OHRMTVXNJQP3895M, OTQRUBEYV0BESSYT, GVYYLXJD6KDARNF4