Getting Started

Getting Started

Get Token Optimizer running in Claude in under 5 minutes.

---

Step 1 — Star the Repository

Required by license before any use.

⭐ Star here

---

Step 2 — Download SKILL.md

Go to the repository and download SKILL.md — this is the only file Claude needs to load all optimization techniques.

Direct link:

https://github.com/samibajwaisking/token-optimizer-skill/blob/main/SKILL.md

---

Step 3 — Load into Claude

Option A — Claude.ai Projects (Recommended)

Best for: daily use, personal workflows, content creation

1. Go to claude.ai
2. Click Projects in the left sidebar
3. Create a new project or open an existing one
4. Click Add content → Upload file
5. Upload SKILL.md
6. Claude now applies all optimization techniques automatically

Test it: Type optimize: write a blog post about AI — Claude will respond using compressed, structured output immediately.

Option B — Paste as System Prompt

Best for: one-time sessions, testing, quick use

1. Open claude.ai → New conversation
2. Click the system prompt field (if available on your plan)
3. Paste the full contents of SKILL.md
4. Start prompting

Option C — API Integration

Best for: developers, automation, production applications

import anthropic

# Load skill
with open("SKILL.md", "r") as f:
    skill = f.read()

client = anthropic.Anthropic(api_key="your-api-key")

response = client.messages.create(
    model="claude-sonnet-4-20250514",
    max_tokens=1024,
    system=skill,
    messages=[
        {"role": "user", "content": "debug+fix: [your code here]"}
    ]
)

print(response.content[0].text)

Option D — Claude Code (CLI)

Best for: developers using Claude in terminal

# Add SKILL.md content to your CLAUDE.md file
cat SKILL.md >> CLAUDE.md

---

Step 4 — Your First Optimized Prompt

Try this immediately after loading the skill:

Unoptimized (old way):

Can you please look at this Python code and tell me what's wrong with it
and how I can fix it to make it work properly?
[code]

Optimized (with this skill):

debug+fix: [code]

Both produce the same quality output. The optimized version uses 85% fewer tokens.

---

Step 5 — Explore the Reference Files

For deeper optimization, load these additional files into your Claude Project:

File	What It Adds
references/compression-patterns.md	50+ shorthand patterns by category
references/output-formats.md	Format templates for every output type
references/system-prompt-templates.md	10 ready system prompts
references/one-shot-examples.md	Examples for 10 domains
references/context-management.md	Conversation pruning techniques

---

Verify It Is Working

After loading the skill, ask Claude:

What token optimization techniques are you currently using?

Claude should respond by listing the 5 pillars. If it does, the skill is active.