🎛️ Complete reference for all CLI flags and options.
ReasonKit’s CLI is designed for power users and automation. Every option has both a short and long form.
These options work with all commands:
| Short | Long | Default | Description |
-v | --verbose | 0 (warn) | Increase logging verbosity (-v info, -vv debug) |
-c | --config | ~/.config/reasonkit/config.toml | Config file path |
-d | --data-dir | ./data | Data directory path |
-h | --help | - | Show help message |
-V | --version | - | Show version information |
| Short | Long | Description |
-p | --protocol <NAME> | Specific protocol to execute (gigathink, laserlogic, etc.) |
| --profile <NAME> | Execution profile (quick, balanced, deep, paranoid) |
-b | --budget <BUDGET> | Adaptive compute budget (e.g., “30s”, “5m”, “$0.50”) |
| --mock | Use mock LLM (no API calls) |
| Short | Long | Default | Description |
| --provider <NAME> | anthropic | LLM provider to use |
-m | --model <NAME> | (Provider default) | Specific model ID |
-t | --temperature <FLOAT> | 0.7 | Generation temperature (0.0-2.0) |
| --max-tokens <INT> | 2000 | Maximum tokens to generate |
| Short | Long | Default | Description |
-f | --format <FORMAT> | text | Output format (text, json) |
| --save-trace | false | Save execution trace to disk |
| --trace-dir <DIR> | | Directory to save traces |
| --list | | List available protocols and profiles |
| Short | Long | Default | Description |
-d | --depth <DEPTH> | standard | Depth of research (quick, standard, deep, exhaustive) |
| --web <BOOL> | true | Include web search results |
| --kb <BOOL> | true | Include knowledge base results |
| --provider <NAME> | anthropic | LLM provider |
-f | --format <FORMAT> | text | Output format (text, json, markdown) |
-o | --output <FILE> | | Save research report to file |
| Short | Long | Default | Description |
-s | --sources <INT> | 3 | Minimum number of independent sources required |
| --web <BOOL> | true | Include web search for verification |
| --kb <BOOL> | true | Include knowledge base sources |
| --anchor | false | Anchor verified content to ProofLedger |
-f | --format <FORMAT> | text | Output format (text, json, markdown) |
-o | --output <FILE> | | Save verification report to file |
Most options can be set via environment variables. See Environment Variables for details.
Options are applied in this order (later overrides earlier):
- Built-in defaults
- Config file settings (
REASONKIT_CONFIG)
- Environment variables
- Command-line flags
# Config says balanced, but CLI overrides to deep
rk think "question" --profile deep