@mandate.md/cli

Policy guardrails for agent wallets, from the command line. Zero install.

npx @mandate.md/cli <command>

Commands

Scan for unprotected calls

Detect wallet calls in your codebase that bypass Mandate:

npx @mandate.md/cli scan

Scans the current directory by default. Pass a path to scan a specific folder:

npx @mandate.md/cli scan ./src

Register

npx @mandate.md/cli login --name "MyAgent"

Stores credentials in ~/.mandate/credentials.json. Prints claimUrl for the owner to link the agent to their dashboard.

Validate (preflight, recommended)

Action-based validation. No gas params, no intentHash.

npx @mandate.md/cli validate \
  --action transfer \
  --amount 10 \
  --to 0xRecipient \
  --token USDC \
  --reason "Invoice #127"

Transfer

ERC20 transfer with policy enforcement:

npx @mandate.md/cli transfer \
  --to 0xAlice \
  --amount 10 \
  --reason "Invoice #127"

Status

npx @mandate.md/cli status <intentId>

Other

npx @mandate.md/cli whoami          # Verify credentials
npx @mandate.md/cli approve <id>    # Wait for human approval
npx @mandate.md/cli activate 0x...  # Set wallet address post-registration

The --raw flag is available on validate for legacy EVM validation with full gas params and intentHash computation.

Agent Discovery

npx @mandate.md/cli --llms          # Machine-readable command manifest
npx @mandate.md/cli --llms-full     # Full manifest with schemas

Agents call --llms to discover commands. Each response includes a next field pointing to the logical next step.

MCP Server

npx @mandate.md/cli --mcp    # Start as MCP stdio server

Community

• Telegram Developer Chat
• Documentation
• GitHub