token-efficiency · delphine-l — Claude Code skill: what it does, how to install
Sets token-saving habits for Claude Code by prioritizing bash filtering, minimal terminal outputs, and targeted file reads.
npx skills add Delphine-L/claude_global --skill token-efficiency -a claude-codeWhat token-efficiency does
The skill embeds practical token-saving guardrails into the agent's workflow. Instead of loading whole files into context, the agent defaults to tools like grep, tail, wc -l, or jq. For simple line substitutions, appends, and file copies, the instructions require direct commands such as sed and cp instead of reading text into memory.
It outlines a model selection pattern: use Opus for initial repository architecture review, then switch to Sonnet for development and debugging. This model split cuts token expenses by around 50% compared to running Opus continuously.
The bash-first rule intentionally yields when editing source code like .py, .js, or .tsx. Whenever diff clarity matters or syntax validation is crucial, the agent switches back to standard file read and edit tools.
What it helps with
- cut token burn during codebase exploration
- inspect structured data and logs using CLI filters instead of context dumps
- split workloads between Opus and Sonnet to reduce spending
- suppress noisy terminal outputs with quiet flags
Example prompts for the agent
- Check app.log for recent errors without dumping the full file
- Inspect metadata in data.json using bash tools
- Summarize the architecture of src under token-efficiency guidelines
How to install token-efficiency
1Skills CLI (Vercel)
npx skills add Delphine-L/claude_global --skill token-efficiency -a claude-code2Manually: put the files in place
~/.claude/skills/token-efficiency/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/token-efficiency/
1Skills CLI (Vercel)
npx skills add Delphine-L/claude_global --skill token-efficiency -a codex2Manually: put the files in place
.agents/skills/token-efficiency/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add Delphine-L/claude_global --skill token-efficiency -a cursor2Manually: put the files in place
.cursor/skills/token-efficiency/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install Delphine-L/claude_global token-efficiency2Manually: put the files in place
.github/skills/token-efficiency/SKILL.md1Skills CLI (Vercel)
npx skills add Delphine-L/claude_global --skill token-efficiency -a gemini-cli2Platform CLI
gemini skills install https://github.com/Delphine-L/claude_global --path skills/claude-meta/token-efficiencyManually: put the files in place
.devin/skills/token-efficiency/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add Delphine-L/claude_global --skill token-efficiency -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about token-efficiency
How do I install token-efficiency in Claude Code?
Run npx skills add Delphine-L/claude_global --skill token-efficiency -a claude-code in your terminal. Alternatively, place the skill into ~/.claude/skills/token-efficiency/SKILL.md.
What does the token-efficiency skill do?
It instructs the agent to save tokens by suppressing command verbosity, searching files with grep, and manipulating files via shell commands without loading them into memory.
When does the skill permit reading files instead of using bash?
The skill instructs the agent to use standard read and edit tools on code files such as .py or .tsx. This keeps diffs reviewable and avoids syntax corruption from blind stream editing.
Similar skills
All skills for Agent setup: memory, context, orchestration →token-efficiency by other authors comes first.
