Skip to content

flags.md

Latest commit

 

History

History
96 lines (69 loc) · 3.76 KB

flags.md

File metadata and controls

96 lines (69 loc) · 3.76 KB

CLI Flags Reference

Flags can be combined freely in any order. Run free-coding-models --help to print this list in-app.

🚀 Tool Launchers

Start the TUI pre-configured to a specific tool. Press Enter on a model to auto-configure and launch it.

Flag Tool What happens on Enter
(none) OpenCode CLI Writes model to opencode.json and launches opencode CLI
--opencode OpenCode CLI Same as above, explicit
--opencode-desktop OpenCode Desktop Writes model to opencode.json and opens desktop app
--openclaw OpenClaw Writes model as primary in ~/.openclaw/openclaw.json
--crush Crush Writes model to ~/.config/crush/crush.json and launches crush
--goose Goose Writes provider config to ~/.config/goose/ and launches goose
--aider Aider Writes model to ~/.aider.conf.yml and launches aider
--qwen Qwen Code Writes model to ~/.qwen/settings.json and launches qwen
--openhands OpenHands Sets LLM_MODEL env var and launches OpenHands
--amp Amp Writes model to ~/.config/amp/settings.json and launches amp
--pi Pi Writes model to ~/.pi/agent/settings.json and launches pi

🔍 Filtering & Display

Flag Type Description
--best boolean Show only top‑tier models (A+, S, S+).
--premium boolean Start with an elite-focused preset (tier filter S + verdict sort). This is fully resettable in the TUI.
--tier <S|A|B|C> value Filter by tier family — S = S+/S, A = A+/A/A-, B = B+/B, C = C only.
--origin <provider> value Filter by provider name (e.g. nvidia, groq, cerebras).
--hide-unconfigured boolean Hide models whose provider has no API key configured.
--show-unconfigured boolean Show all models regardless of API key configuration (overrides default).

📊 Sorting

Flag Type Description
--sort <column> value Start sorted by a specific column. Valid values: rank, tier, origin, model, ping, avg, swe, ctx, condition, verdict, uptime, stability.
--asc boolean Sort ascending (smallest first).
--desc boolean Sort descending (largest first).

📤 Output Modes

Flag Type Description
--json boolean Skip the TUI — print all model results as a JSON array and exit. Combine with jq for scripting.
--fiable boolean Wait 10 s, pick the most reliable model by avg + stability + uptime, print provider/model_id and exit.
--recommend boolean Open the Smart Recommend overlay immediately on startup (same as pressing Q).

⚙️ Runtime Options

Flag Type Description
--ping-interval <ms> value Override the ping interval in milliseconds (e.g. --ping-interval 5000).
--no-telemetry boolean Disable anonymous usage telemetry for this session.
--help, -h boolean Print the full help text with all flags and exit.

Examples

# Start in Crush mode filtered to S-tier only
free-coding-models --crush --tier S

# Get the fastest S-tier model ID (headless, for scripts)
free-coding-models --tier S --json | jq -r '.[0].modelId'

# Filter by latency
free-coding-models --json | jq '.[] | select(.avgPing < 500)'

# Most reliable model right now
free-coding-models --fiable

# Start with an elite-focused preset (resettable in-app)
free-coding-models --premium

# Sort by SWE score descending
free-coding-models --sort swe --desc

# Groq models only
free-coding-models --origin groq

# Configure Goose with an S-tier model
free-coding-models --goose --tier S