Commands Reference

Complete reference for all LynxPrompt CLI commands, options, and flags.

Quick Reference

Command	Description
`lynxp login`	Authenticate with LynxPrompt
`lynxp wizard`	Generate AI config interactively (recommended)
`lynxp analyze`	Analyze project tech stack (local or remote)
`lynxp convert`	Convert between config formats
`lynxp merge`	Merge multiple config files
`lynxp import`	Scan repo for AGENTS.md files and AI commands
`lynxp check`	Validate AI config files (CI/CD)
`lynxp status`	Show config status and tracked blueprints
`lynxp pull <id>`	Download and track a blueprint
`lynxp push`	Push local file or command to cloud
`lynxp search <query>`	Search marketplace blueprints
`lynxp list`	List your blueprints
`lynxp link`	Link local file to cloud blueprint (interactive)
`lynxp unlink`	Disconnect file from cloud blueprint (interactive)
`lynxp diff`	Compare all tracked files with cloud
`lynxp sync`	Pull cloud changes and push local ones for all tracked files
`lynxp config`	Show or change CLI settings (API URL, config path)

Note: lynxp is an alias for lynxprompt.

lynxp wizard

The recommended way to generate AI configuration files. Interactive wizard that creates AGENTS.md, .cursor/rules/, or any other format based on your choices.

lynxp wizard [options]

Basic Options:
  -n, --name <name>         Project name
  -d, --description <desc>  Project description
  -s, --stack <stack>       Tech stack (comma-separated)
  -f, --format <format>     Output: agents, cursor, copilot, etc.
  --persona <persona>       AI persona (fullstack, backend, frontend, etc.)
  --boundaries <level>      Boundary preset (conservative, standard, permissive)
  -y, --yes                 Skip prompts, use defaults

Advanced Options:
  -o, --output <dir>        Output directory
  --repo-url <url>          Analyze remote repository (GitHub/GitLab)
  --blueprint               Generate with [[VAR|default]] placeholders
  --license <type>          License type (mit, apache-2.0, gpl-3.0)
  --ci-cd <platform>        CI/CD platform (github_actions, gitlab_ci)
  --project-type <type>     Project type (work, leisure, opensource)
  --detect-only             Only detect project info, don't generate
  --load-draft <name>       Load a saved wizard draft
  --save-draft <name>       Save wizard state as draft
  --vars <values>           Fill variables: VAR1=val1,VAR2=val2

Examples

# Interactive wizard (recommended)
$ lynxp wizard

# Quick generation with defaults
$ lynxp wizard -y

# Generate for Cursor
$ lynxp wizard -f cursor

# Analyze remote repo first
$ lynxp wizard --repo-url https://github.com/owner/repo

# Generate blueprint template
$ lynxp wizard --blueprint
# Output: [[PROJECT_NAME|MyApp]] configuration...

# Fill blueprint variables
$ lynxp wizard --blueprint --vars "PROJECT_NAME=MyApp,LICENSE=MIT"

# Save/load drafts
$ lynxp wizard --save-draft myproject
$ lynxp wizard --load-draft myproject

Remote Repository Detection

Analyze remote repositories via GitHub or GitLab API to auto-detect tech stack, databases, CI/CD, Docker, and more. For other hosts, a shallow clone is used automatically.

lynxp analyze

Analyze project configuration and tech stack without generating files. Supports local directories and remote repositories (GitHub, GitLab).

lynxp analyze [options]

Options:
  -r, --remote <url>    Analyze remote repository
  -j, --json            Output as JSON (for scripting)

Examples

# Analyze current directory
$ lynxp analyze
📊 Project Analysis
  Name: my-project
  Stack: typescript, react, tailwind
  Package Manager: npm
  CI/CD: github_actions

# Analyze remote repository
$ lynxp analyze -r https://github.com/owner/repo

# JSON output for scripts
$ lynxp analyze --json | jq '.detected.stack'

lynxp convert

Convert AI configuration files between formats. Useful for migrating between editors or generating multiple formats from a single source. The rules are copied as written; only the frontmatter a tool needs (Cursor, Kiro, Augment) is added or removed, the same as the web download. Tools that read a JSON or YAML config instead of a rules file get theirs from lynxp wizard.

lynxp convert [source] <target>

Arguments:
  source    Source file (auto-detected if omitted)
  target    Target format: agents, cursor, copilot, windsurf, etc.

Options:
  -o, --output <file>   Output filename
  -f, --force           Overwrite existing file

Examples

# Convert AGENTS.md to Cursor format
$ lynxp convert AGENTS.md cursor
✓ Converted to .cursor/rules/project.mdc

# Auto-detect source and convert
$ lynxp convert cursor

# Custom output filename
$ lynxp convert AGENTS.md copilot -o team-rules.md

MCP servers

An MCP config converts to any other tool's MCP config: the target is the tool's id followed by -mcp(an unknown target lists them all). Each entry is rewritten in the target's shape: VS Code'sservers, Gemini CLI's httpUrl, OpenCode's command array, Zed'scontext_servers, Codex and Mistral Vibe TOML. When the target is a settings file (.gemini/settings.json,.zed/settings.json, opencode.json,.codex/config.toml...), only its MCP servers change and every other setting stays; lynxp pull andlynxp sync do the same, and lynxp pushsends only the MCP servers of such a file.

# The project's .mcp.json servers for Gemini CLI, other settings kept
$ lynxp convert .mcp.json gemini-mcp

# From the first MCP config found, to Codex
$ lynxp convert codex-mcp

lynxp merge

Merge two or more AI configuration files into one. Useful for combining team rules, project-specific configs, or rules from different sources.

lynxp merge <file1> <file2> [...files]

Options:
  -o, --output <file>      Output filename (default: merged.md)
  -s, --strategy <type>    Merge strategy: concat, sections, smart
  -f, --force              Overwrite existing file
  -i, --interactive        Review and select sections to include

Examples

# Merge two config files
$ lynxp merge team-rules.md project-rules.md
✓ Merged to merged.md

# Custom output and strategy
$ lynxp merge a.md b.md c.md -o combined.md -s smart

# Interactive mode to select sections
$ lynxp merge rules1.md rules2.md -i

Merge Strategies

concat — Simple concatenation with separators
sections — Group by section title
smart — Dedupe similar content (default)

lynxp import

Scan a repository for existing AGENTS.md files, AI commands, and understand your monorepo structure. Perfect for discovering AI configurations and slash commands across large codebases.

lynxp import [path] [options]

Arguments:
  path              Directory to scan (default: current directory)

Options:
  --dry-run         Preview what would be imported
  --no-recursive    Don't scan subdirectories
  --depth <n>       Max directory depth to scan (default: 10)
  --pattern <file>  Custom config filename to look for
  -v, --verbose     Show detailed section information
  -j, --json        Output as JSON (for scripting)

Examples

# Scan current directory
$ lynxp import
📥 LynxPrompt Import
   Found 4 configuration file(s)

🏢 my-monorepo [Monorepo Root]
   AGENTS.md
   └─ 📄 packages/web
      packages/web/AGENTS.md
   └─ 📄 packages/api
      packages/api/AGENTS.md

📊 1 monorepo detected with hierarchical configs

# Preview without changes
$ lynxp import --dry-run

# Scan specific directory
$ lynxp import ./packages --depth 3

# JSON output for scripting
$ lynxp import --json | jq '.hierarchy'

What it detects

📋 AI Rules & Configuration

• AGENTS.md — Universal AI config format
• CLAUDE.md — Claude Code format
• .cursorrules — Legacy Cursor format
• .windsurfrules — Windsurf format

⚡ AI Agent Commands (Slash Commands)

• .cursor/commands/*.md — Cursor commands
• .claude/commands/*.md — Claude Code commands
• .devin/workflows/*.md — Devin Desktop workflows (and the old .windsurf/workflows/)
• .github/prompts/*.prompt.md — GitHub Copilot prompts
• .opencode/commands/*.md — OpenCode commands
• .gemini/commands/*.toml, .kiro/prompts/, .clinerules/workflows/ and the command folders of the other supported tools

Monorepo Support: The import command automatically detects hierarchical configurations where package-level AGENTS.md files inherit from a root configuration. This information is saved to .lynxprompt/hierarchy.json for use by other commands.

lynxp check

Validate AI configuration files. Perfect for CI/CD pipelines and pre-commit hooks.

lynxp check [options]

Options:
  --ci        CI mode - exit codes only (0=pass, 1=fail)
  -j, --json  Output as JSON (for scripting)

Example

$ lynxp check
🐱 LynxPrompt Check

✓ Found 3 configuration files:
  AGENTS.md
  .cursor/rules/project.mdc
  .github/copilot-instructions.md

⚠ 1 warning:
  AGENTS.md: Contains placeholder text "TODO"

✅ Validation passed!

# In CI/CD:
$ lynxp check --ci
✓ Validation passed

What it validates

• Empty or minimal content
• Placeholder text (TODO, FIXME, etc.)
• Potential secrets or API keys
• Hidden Unicode (zero-width, bidi and tag characters) that can carry instructions a reviewer cannot see. A lone direction mark next to plain text, common in right-to-left text, is only a warning
• YAML frontmatter syntax (for .mdc files)
• LynxPrompt config file validity

lynxp pull

Download a blueprint from the marketplace and automatically track it for future updates.

lynxp pull <blueprint-id> [options]

Arguments:
  blueprint-id    The blueprint ID (e.g., bp_abc123)

Options:
  -o, --output <dir>    Output directory (default: current)
  -y, --yes             Skip confirmation prompts
  --preview             Preview content without downloading
  --no-track            Don't track for future syncs
  --vars <values>       Fill [[VARIABLES]]: VAR1=value1,VAR2=value2

Examples

# Download and track
$ lynxp pull bp_abc123
🐱 Blueprint: Next.js TypeScript Starter
   📦 Marketplace blueprint (read-only)

✅ Downloaded: AGENTS.md
   Linked to: bp_abc123
   Updates: Run 'lynxp pull bp_abc123' to sync updates

# Preview first
$ lynxp pull bp_abc123 --preview
# Shows content without writing file

# Without tracking
$ lynxp pull bp_abc123 --no-track

# Fill variables (unset ones get their [[NAME|default]])
$ lynxp pull bp_abc123 --vars PROJECT_NAME=acme,TEAM=Platform

Blueprint Types

Marketplace - Read-only, can pull updates but not push changes
Team - Full sync, push and pull changes
Private - Your own, full control

lynxp link

Connect an existing local file to a cloud blueprint for tracking. Interactive mode guides you through the process when run without arguments.

lynxp link                         # Interactive mode (recommended)
lynxp link <file> <blueprint-id>   # Direct mode
lynxp link --list                  # List tracked blueprints

Interactive Example

$ lynxp link
🐱 Link File to Blueprint

? Which file do you want to link?
  → AGENTS.md
    .cursor/rules/project.mdc

? How do you want to find the blueprint?
  → 📋 From my blueprints
    🔍 Search marketplace
    🔢 Enter ID directly

? Select a blueprint:
  → Next.js TypeScript Starter
    Team Standards

✅ Linked: AGENTS.md → bp_abc123

# Or use direct mode:
$ lynxp link AGENTS.md bp_abc123

lynxp unlink

Disconnect a local file from its cloud blueprint. Interactive mode when run without arguments.

lynxp unlink           # Interactive mode (shows tracked files)
lynxp unlink <file>    # Direct mode

Example

$ lynxp unlink
🐱 Unlink File from Blueprint

? Which file do you want to unlink?
  → AGENTS.md (Next.js TypeScript Starter)
    .cursor/rules/project.mdc (Team Standards)

Currently linked to: Next.js TypeScript Starter
   ID: bp_abc123
   Source: marketplace

? Unlink AGENTS.md? Yes

✅ Unlinked: AGENTS.md

lynxp diff

Compare tracked files with their cloud blueprints. Run without arguments to check all tracked files at once.

lynxp diff                   # Compare ALL tracked files with cloud
lynxp diff <file>            # Compare specific file with its blueprint
lynxp diff <blueprint-id>    # Compare with specific blueprint
lynxp diff --local           # Compare .lynxprompt/rules/ with exports

Examples

# Compare all tracked files (recommended)
$ lynxp diff
🐱 LynxPrompt Diff

📄 AGENTS.md
   Linked to: Next.js Starter (bp_abc123)
   ✓ In sync with cloud

📄 .cursor/rules/project.mdc
   Linked to: Team Standards (bp_team456)
- Old line
+ New line
   +2 -1 lines

To push local changes: lynxp push

# Compare specific file
$ lynxp diff AGENTS.md

# Compare with untracked blueprint
$ lynxp diff bp_newid123

lynxp sync

Bring every tracked file and its cloud blueprint together. Each file is compared with the version you last pulled or pushed: cloud changes are pulled, local changes are pushed, and when both sides changed the cloud version is written next to your file as <file>.remote so nothing is overwritten.

lynxp sync               # Sync all tracked files
lynxp sync <file>        # Sync one tracked file
lynxp sync --dry-run     # Show what would happen, change nothing

Example

$ lynxp sync
🐱 LynxPrompt Sync

↓ AGENTS.md: updated from the cloud
↑ CLAUDE.md: pushed local changes
✗ .cursor/rules/project.mdc: changed both locally and in the cloud
   Cloud version written to .cursor/rules/project.mdc.remote; .cursor/rules/project.mdc was not touched.
   Merge what you need into .cursor/rules/project.mdc, delete .cursor/rules/project.mdc.remote, then run 'lynxp sync' again.

1 pulled, 1 pushed, 0 in sync, 1 conflict

Marketplace blueprints are read-only: local edits stay local and are never pushed.
A file pulled with --vars gets cloud updates filled with the same values, and is not pushed back.
While <file>.remote exists, sync and push leave that file alone.
Exits with code 1 when a file has a conflict or could not be synced.

lynxp status

Show current AI configuration status including tracked blueprints and local files.

lynxp status

Example output

$ lynxp status

🐱 LynxPrompt Status
   Directory: /Users/me/my-project

✓ LynxPrompt initialized
   Exporters: cursor, agents

📦 Tracked Blueprints

  ✓ AGENTS.md [marketplace]
     ID: bp_abc123 • Next.js Starter

  ● .cursor/rules/project.mdc [team]
     ID: bp_team456 • Team Standards
     ⚠ Local changes - run 'lynxp push' to sync

📄 AI Config Files

  ✓ AGENTS.md (tracked)
     Platform: Claude Code, Cursor, AI Agents
     Size: 2.1 KB (85 lines)

  ✓ .cursor/rules/ (tracked)
     Platform: Cursor
     Rules: 3 files

  ✓ .github/copilot-instructions.md
     Platform: GitHub Copilot
     Size: 1.2 KB (42 lines)

lynxp search

Search public blueprints in the LynxPrompt marketplace. No authentication required.

lynxp search <query> [options]

Arguments:
  query           Search terms

Options:
  --limit <n>     Maximum results (default: 20)
  -j, --json      Output as JSON (for scripting)

Authentication Commands

lynxp login

Opens browser to authenticate

lynxp logout

Remove stored credentials

lynxp whoami

Show current user

lynxp config

View and manage CLI configuration. Point the CLI to a self-hosted instance or inspect the current settings.

lynxp config                    # Show current settings
lynxp config set-url <url>      # Point to a self-hosted instance
lynxp config reset-url          # Reset to default (lynxprompt.com)
lynxp config path               # Show config file location

Examples

# View current configuration
$ lynxp config
⚙️  LynxPrompt CLI Configuration
  API URL: https://lynxprompt.com
  Login:   user@example.com

# Point to self-hosted instance (each instance keeps its own login)
$ lynxp config set-url https://lynxprompt.yourcompany.com
✓ API URL set to https://lynxprompt.yourcompany.com
  No saved login for this instance. Run 'lynxp login' to authenticate with it.

# Reset to default (a saved login is checked with the instance)
$ lynxp config reset-url
✓ API URL reset to https://lynxprompt.com
  Logged in as user@example.com on this instance (token checked).

# Find config file
$ lynxp config path
~/.config/lynxprompt/config.json

Config precedence

The LYNXPROMPT_API_URL environment variable takes precedence over the stored config file value. This is useful for CI/CD pipelines where you want to temporarily override the URL.

Environment Variables

Variable	Description
`LYNXPROMPT_TOKEN`	API token for CI/CD (skips browser auth)
`LYNXPROMPT_API_URL`	Custom API URL (default: https://lynxprompt.com)

Need more help?

Check the FAQ or reach out to our support team.

Troubleshooting