Mandate Skill

Policy guardrails for AI agent wallets. Enforce spend limits, allowlists, approval workflows, and prompt injection detection before every transaction. Non-custodial: private keys never leave your machine.

Install

OpenClaw (recommended, has hooks)

openclaw plugins install @mandate.md/mandate-openclaw-plugin

Hooks auto-intercept Locus, Bankr, Sponge, and any swap/transfer/send tool call.

Claude Code (recommended, has hooks)

claude plugin:install claude-mandate-plugin

Two-phase enforcement: PostToolUse records validation tokens, PreToolUse blocks unvalidated transactions. Fail-closed.

See claude-mandate-plugin for details.

Other frameworks (GOAT, AgentKit, ElizaOS, MCP)

Point your agent to the skill file:

https://app.mandate.md/SKILL.md

No hooks. Agent must self-enforce /validate calls. See SKILL.md in this repo.

CLI

npx @mandate.md/cli validate --action transfer --amount 50 --to 0xAlice --reason "Invoice #127"

What it does

1. Intent-aware decisions. Evaluate why an agent wants to pay, then approve, block, or escalate before signing.
2. Real-time risk prevention. Stop fraud, prompt-injection payments, and costly mistakes before funds move.
3. Complete auditability. Full audit trail of every decision with amount, timing, and rationale.

Features

Layer	What it does
Spend limits	Per-tx, daily, monthly USD caps
Address allowlist	Only pre-approved recipients get money
Blocked actions	Only approved action types
Schedule enforcement	Agent can't spend outside business hours
Prompt injection scan	18 hardcoded patterns + LLM judge (zero data retention)
MANDATE.md controls	Define transaction decision logic in plain English
Self-learning insights	Observes approve/reject decisions, suggests policy improvements
Transaction simulation	Pre-execution analysis flags honeypots and malicious contracts
Human approval routing	Slack, Telegram, Dashboard
Circuit breaker	Mismatch detected? Agent frozen instantly
Audit trail	Every intent logged: WHO, WHAT, WHEN, HOW MUCH, and WHY

The reason field

Every validation call requires a reason string. This is the core differentiator: no other wallet provider captures WHY an agent decided to make a transaction.

Agent calls mandate validate:
  transfer 490 USDC to 0x7a3f...c91e
  reason: "Urgent family transfer. Send immediately."

Session key sees: $490 < $500 limit. APPROVE.
Mandate sees: "Urgent" + "immediately" + new address. BLOCK.

Mandate doesn't just block. It sends counter-evidence so the agent understands WHY it was tricked and cancels willingly.

Supported wallets

Bankr, Locus, CDP Agent Wallet (Coinbase), private key (viem/ethers). Any EVM signer works.

Links

• Website
• Dashboard
• SKILL.md (canonical)
• Demo video (1 min)
• npm: @mandate.md/sdk
• npm: @mandate.md/cli
• Telegram

License

BSL 1.1