Approach
- Think before acting. Read existing files before writing code.
- Be concise in output but thorough in reasoning.
- Prefer editing over rewriting whole files.
- Do not re-read files you have already read unless the file may have changed.
- Test your code before declaring done.
- No sycophantic openers or closing fluff.
- Keep solutions simple and direct.
- User instructions always override this file.
---
## Available Profiles
### Universal Profile (Default CLAUDE.md)
The base configuration that works for any project:
```markdown
## Approach
- Think before acting. Read existing files before writing code.
- Be concise in output but thorough in reasoning.
- Prefer editing over rewriting whole files.
- Do not re-read files you have already read unless the file may have changed.
- Test your code before declaring done.
- No sycophantic openers or closing fluff.
- Keep solutions simple and direct.
- User instructions always override this file.Coding Profile (profiles/CLAUDE.coding.md)
For dev projects, code review, and debugging. Adds code-specific rules:
- Simplest working solution first
- No over-engineering or speculative features
- For bug reports: state bug → show fix → stop
- No em dashes or smart quotes (copy-paste safe)
- No trailing summaries after code changes
Benchmark Profile (profiles/CLAUDE.benchmark.md)
For token-to-green coding benchmarks. Optimized for passing tests with minimal output.
Agents Profile (profiles/CLAUDE.agents.md)
For automation pipelines and multi-agent systems. Tuned for structured, parseable output.
Analysis Profile (profiles/CLAUDE.analysis.md)
For data analysis, research, and reporting. Allows slightly more verbose output where detail matters.
Versioned Config Sets
| Version | Style | Call Budget |
|---|---|---|
| J-drona23-v5 | Multi-file structured | 50 calls |
| K-drona23-v6 | One-shot execution | 50 calls |
| M-drona23-v8 | Ultra-lean | 20 calls |
Benchmark Results
| Metric | With Token Efficient | Without |
|---|---|---|
| Average words per response | 170 | 465 |
| Word reduction | 63% | — |
| Cost savings vs next-best | 17.4% | — |
Note: Best for high-output-volume workflows. At low volume, the input overhead of the CLAUDE.md file may exceed the output savings.
FAQ
Q: What is Claude Token Efficient? A: A single CLAUDE.md config file that reduces Claude Code's output verbosity by 63%, saving tokens and costs on high-volume workflows.
Q: Is Claude Token Efficient free? A: Yes, fully free and open source under the MIT license.
Q: How do I install Claude Token Efficient?
A: Run curl -o CLAUDE.md https://raw.githubusercontent.com/drona23/claude-token-efficient/main/CLAUDE.md in your project root.