The jfp CLI is designed for speed and simplicity. Here are the essential commands.
Quick start
Running jfp with no arguments shows a quick help guide:
$ jfp JeffreysPrompts CLI - Access battle-tested prompts from your terminal USAGE: jfp [command] [options] COMMANDS: list List all prompts show <id> Show a specific prompt by ID search <query> Search prompts recommend Get prompt recommendations impact <id> Show downstream dependencies graph export Export dependency graph categories List all categories export Export prompts to files i Interactive browser FLAGS: --json Output as JSON --raw Output raw prompt content --help, -h Show help --version, -v Show version
Free vs Pro
The CLI is free to build and use with the public prompt library. Pro unlocks your private vault and premium content.
- Free: list/search/show/copy/render/export public prompts, bundles, and JSON output
- Pro: personal prompts (
list --mine,list --saved,search --mine,search --saved,search --all), plussave,sync,notes,collections,packs(premium pack installs),cost(token/cost estimates), and metadata tools (tags suggest,dedupe scan)
If you log in on a free plan, Pro-only flags will prompt you to upgrade.
Learn more about Pro at pro.jeffreysprompts.com/pricing.
Listing prompts
# List all prompts jfp list # List prompts in a category jfp list --category debugging # List with JSON output jfp list --json
Recommendations
Use recommend to discover prompts you may like. If you have synced your saved prompts, the results are personalized; otherwise you'll see featured prompts.
# Personalized recommendations (uses saved prompts when available) jfp recommend # Related recommendations for a specific prompt jfp recommend idea-wizard
Cost estimates (Pro)
Estimate token usage and USD cost per prompt. Cost estimates require a Pro subscription.
# Estimate cost for a prompt jfp cost idea-wizard # Override model and output tokens jfp cost idea-wizard --model gpt-4o-mini --output-tokens 500 # List supported pricing models jfp cost --list-models # View recent budget alerts jfp cost --alerts --alerts-limit 10
Metadata tools (Pro)
Use the metadata assistant to suggest tags or find likely duplicate prompts.
# Suggest tags/categories/descriptions for a prompt jfp tags suggest idea-wizard # Scan for near-duplicate prompts jfp dedupe scan --min-score 0.9 --limit 10
Impact analysis
Use impact to see what depends on a prompt, or export the full dependency graph. You can also include category/tag metadata or your collections as graph nodes.
# Show downstream dependencies jfp impact idea-wizard # Export dependency graph as JSON jfp graph export --json # Include categories and tags jfp graph export --include-meta --json # Include collections (requires login) jfp graph export --include-collections --json
Getting a specific prompt
Use the show command with a prompt ID:
# Get a prompt by ID jfp show idea-wizard # Output as JSON jfp show idea-wizard --json # Copy directly to clipboard (macOS) jfp show idea-wizard --raw | pbcopy
Viewing categories
# List all categories jfp categories # Categories include prompt counts by default
Output formats
jfp supports multiple output formats:
- Default β Human-readable with colors
--jsonβ Machine-readable JSON--rawβ Just the prompt content, no formattingexport --format mdβ Markdown export
# JSON for scripting jfp show code-reviewer --json # Raw prompt for piping jfp show code-reviewer --raw | pbcopy # Markdown for documentation jfp export code-reviewer --format md --stdout > prompt.md
Using with AI coding agents
jfp is optimized for use with AI coding agents like Claude Code:
# Get a prompt and pipe to an agent jfp show code-reviewer --raw | claude # Use JSON output for structured parsing jfp search "debugging" --json --limit 5 | jq -r '.results[0].id'
Common options
| Option | Description |
|---|---|
--json | Output as JSON |
--raw | Just the content, no metadata |
--category <cat> | Filter list by category |
--limit <n> | Limit results (search/suggest) |
JFP_NO_COLOR=1 | Disable colored output (env var) |
Exit codes
jfp uses standard exit codes:
0β Success1β Error (most failures)130β User cancelled an interactive prompt