{"version":"2.0.46","anchor":"effort-level-configuration","canonical_anchor":"effort-level-configuration","heading":"Effort Level Configuration","tier":null,"area":null,"scope":null,"heads_up":null,"url":"https:\/\/changelogs.core-directive.com\/v\/2.0.46\/e\/effort-level-configuration","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.0.46","markdown":"### Effort Level Configuration\n\n**What:** Control Claude's reasoning effort\/intensity for tasks through multiple configuration methods\n\n**How to use:**\n\n`# Via environment variable export CLAUDE_CODE_EFFORT_LEVEL=high claude # Supported values: \"low\", \"medium\", \"high\", or numeric values export CLAUDE_CODE_EFFORT_LEVEL=75 claude # Disable with \"unset\" export CLAUDE_CODE_EFFORT_LEVEL=unset claude` **Details:**\n\n- Configuration priority: CLI flags (highest) \u2192 environment variable \u2192 settings file (lowest)\n- Set via `CLAUDE_CODE_EFFORT_LEVEL` environment variable\n- Configure in settings file with `effortLevel` field\n- Supports string presets: \"low\", \"medium\", \"high\"\n- Supports numeric values (integers)\n- Use \"unset\" to explicitly disable effort configuration\n- Integrates with Claude API's `output_config.effort` parameter using beta flag `effort-2025-11-24`\n- **Evidence**: New function `a10() at line 341081` retrieves configuration from multiple sources; function `nR3() at line 491217` sets API parameters; replaces empty stub `uR3() at line 491090` from v2.0.45"}