| name | Tune Settings |
| description | This skill should be used when the user says "optimize settings.json", "tune settings for low tokens", "settings.json for performance", "disable auto memory", "configure for token savings", "minimal context window settings", or "settings taking too many tokens". |
| version | 1.0.0 |
Tune Settings
Read and optimize settings.json (or settings.local.json) for minimal token usage. Show diff before writing.
Step 1: Find Settings Files
Check in order:
./.claude/settings.json (project)
./.claude/settings.local.json (project local)
~/.claude/settings.json (global)
~/.claude/settings.local.json (global local)
Read all found files.
Step 2: Analyze Current Settings
Identify token-costly settings:
| Setting | Cost | Recommendation |
|---|
autoLoadMemory: true | High | Set to false |
autoLoadSkills: true | Very High | Set to false |
| Verbose logging enabled | Medium | Disable |
Large contextWindow | High | Reduce to needed size |
| Hook scripts that read files | Medium | Audit hooks |
| Multiple MCP servers active | Medium | Disable unused |
Step 3: Apply Token-Saving Config
Recommended low-token settings.json:
{
"autoLoadMemory": false,
"autoLoadSkills": false,
"compactOnContextFull": true,
"verboseOutput": false,
"plugins": {
"autoEnable": false
}
}
Key settings explained:
autoLoadMemory: false — Prevents .remember/ files from loading every session
autoLoadSkills: false — Skills load only when explicitly triggered
compactOnContextFull: true — Auto-compact instead of erroring at context limit
plugins.autoEnable: false — Plugins require explicit activation
Step 4: Show Diff and Confirm
Display before/after diff:
- "autoLoadMemory": true,
+ "autoLoadMemory": false,
- "autoLoadSkills": true,
+ "autoLoadSkills": false,
+ "compactOnContextFull": true,
Ask user to confirm scope: project settings or global settings.
Step 5: Write File
On confirmation, write the updated settings file.
Warning: Changing global settings affects all Claude Code sessions.
LTX Schema
Emit structured output as LTX rows when reporting settings analysis results.
@v1:setting|current|recommended|action
| Field | Description |
|---|
setting | Settings key (e.g. autoLoadMemory) |
current | Current value or missing if not set |
recommended | Recommended value for token savings |
action | change, keep, add |
Example:
@v1:setting|current|recommended|action
autoLoadMemory|true|false|change
autoLoadSkills|true|false|change
compactOnContextFull|missing|true|add
verboseOutput|false|false|keep
Additional Resources
references/settings-reference.md — Full settings.json reference with token impact ratings