CLI

Peregrine includes a command-line interface for security workflows and for launching the interactive terminal experience.

Run direct commands for scripted analysis or start the full TUI to work interactively with agents.

Interactive interface

Run peregrine to launch the TUI workbench.

The workbench layout includes:

• Explorer pane on the left for file navigation.
• Main workspace with chat, editor, bytecode, and graphs tabs.
• Status bar at the bottom displaying build, test, fuzz, and verification metrics.

Open a specific project by passing a path:

peregrine /path/to/your/project

The TUI provides slash commands, agent interactions, and the same capabilities as the desktop app.

Agent mode

Start the agent directly with:

$ peregrine agent

Configure provider API keys with /provider or run local models via Ollama/LM Studio. Pass model overrides and other configuration for scripted or embedded agent runs.

Direct commands

Target a Move package and run analysis commands.

Use global flags to specify the project and package:

• --project <path> (defaults to current directory)
• --package <name> (defaults to current directory)
• --json for machine-readable output

Security analysis

Common commands include:

• build — Build the package
• test — Run tests
• coverage — Generate coverage data
• bytecode (or bytecode-viewer) — Inspect compiled bytecode and control flow
• signatures (or function-signatures) — List public and entry function signatures
• call-graph — Generate call graphs
• object-graph — Generate object graphs
• cfg (or control-flow-graph) — Generate control-flow graphs
• fuzz — Run fuzzing against target functions
• verify — Perform formal verification on modules
• analyze — Run advanced multi-stage analysis
• check-all — Execute a comprehensive set of checks (with flags like --skip-fuzz)

Package management

• new-package — Create a new boilerplate Move package
• import-package — Import an on-chain package by ID and network

These commands let you bring code into your environment for analysis.

Output and scripting

Add --json to commands to get structured output suitable for CI, scripts, or other tools.

The CLI shares the same model-agnostic and autonomous agent foundation as the rest of Peregrine.

Next steps

See the Audit workflow page for how to run full audits from the command line or TUI.

See Subagents for details on agent collaboration.