Bank of Bots

Trust scoring for AI agents. Log transactions and submit payment proofs to build a verifiable BOB Score — a trust score (think FICO but for AI Agents) that o...

Security Scan

The skill's instructions let an agent perform and configure real financial operations but the registry metadata omits the sensitive credentials and config changes the SKILL.md requires — the pieces don't fully line up, so proceed cautiously.

Purpose & Capability

The name/description emphasize 'trust scoring' (logging and proof submission), but the runtime instructions include full wallet and payment management (funding wallets, executing sends, ACH/wire, counterparty management, operator-only actions). That broader financial capability could be legitimate for a payment-linked trust system, but it's a substantial expansion of scope compared with a pure 'scoring' skill and should be justified by an authoritative source.

Instruction Scope

SKILL.md instructs the agent to call a 'bob' CLI or the Bank of Bots REST API and to run commands that reveal and modify CLI configuration (bob config show / set). Those commands can expose local config paths and let the agent change where it sends API requests. The file explicitly requires using BOB_API_KEY, BOB_AGENT_ID, and BOB_API_URL even though the registry lists no required env vars. The instructions therefore expand scope to accessing secrets and modifying client configuration.

Install Mechanism

This is an instruction-only skill with no install spec and no code files, so nothing is written to disk by the skill itself. That lowers install-time risk.

Credentials

SKILL.md requires sensitive environment values (BOB_API_KEY, BOB_AGENT_ID, BOB_API_URL) and implies operator-level capabilities (funding wallets, executing transfers). The registry metadata declares no required env vars — a clear mismatch. Requesting credentials capable of initiating payments or changing config is high privilege and should be explicitly declared and minimized.

Persistence & Privilege

The skill is not always-on and has no install hooks, which is good. However the instructions include commands that modify the CLI's API URL (bob config set api-url), meaning an agent following these instructions could redirect requests to another endpoint if given permission to run those commands. Autonomous invocation is allowed by policy default; combined with the above environment/credential issues this increases risk and warrants caution.

What to consider before installing

Do not install or run this skill until you verify its provenance and intent. Specific actionable checks: (1) Ask the publisher for an official homepage or source repo and confirm the published SKILL.md matches the official docs. (2) Treat BOB_API_KEY and BOB_AGENT_ID as highly sensitive — don't paste operator or funding credentials unless you trust the service and have verified the code. (3) If you test, use a low-privilege agent/account with no custody or funding rights and a non-production BOB_API_URL to avoid accidental payments. (4) Be cautious about allowing the agent to run 'bob config set' or similar commands that can change API endpoints. (5) Prefer an explicit registry declaration of required env vars and required roles; absence of those in the metadata is a red flag. If you need a stronger assessment, provide the skill's publisher/homepage or the actual bankofbots API documentation so I can reconcile the intended scope with the runtime instructions.

Setup

If you haven't connected yet, follow the setup guide: https://app.bankofbots.ai/docs/agent-setup

Once connected, you need these credentials for all API calls:

• BOB_API_KEY — your agent API key
• BOB_AGENT_ID — your agent ID
• BOB_API_URL — https://api.bankofbots.ai/api/v1

You can use the bob CLI (optional) or call the REST API directly. All commands output structured JSON with an ok field and next_actions that suggest follow-up commands.

Non-custodial v0 mode note: use BTC rails (lightning, onchain) and proof flows. Direct custodial ingress endpoints (wallet fund, wallet receive, receive confirm, operator fund) return 409 Conflict.

Core concepts

• Agent: An AI agent with its own identity, wallets, and spending policies
• Wallet: Holds a balance in a specific currency (USD, BTC, USDC). Each agent can have multiple wallets.
• Policy: Spend limits, rate limits, and kill switches that constrain agent behavior
• Transaction: A record of money moving in or out of a wallet
• Payment: An outbound payment to an external bank account (ACH, wire) or internal account (book)
• Counterparty: A saved bank account destination for ACH/wire payouts
• Service Gate: A priced action that callers unlock by presenting a completed payment intent

Commands

Check your identity

bob auth me

Returns your role (agent or operator), identity details, and role-aware next_actions.

Agent details and wallet balances

bob agent get <agent-id>

Response includes a wallets array with each wallet's balance, currency, rail, and status.

Wallet management

# List wallets for an agent
bob wallet list <agent-id>

# Fund a wallet (operator-only, legacy custody mode)
bob wallet fund <agent-id> --wallet-id <id> --amount <cents>

# Generate a receive invoice or deposit address
bob wallet receive <agent-id> --rail lightning --amount <sats>
bob wallet receive <agent-id> --rail onchain
# receive is disabled in non-custodial v0 mode (409 Conflict)

# Get/set wallet budget (smallest currency unit)
bob wallet budget get <agent-id> --wallet-id <id>
bob wallet budget set <agent-id> --wallet-id <id> --amount <cents>

bob wallet list now includes a bob_address field on each wallet when a default agent address is available. In non-custodial v0 mode, bob wallet fund and bob wallet receive are intentionally disabled (409 Conflict).

Flag	Description
--rail	Required: lightning or onchain (auto-selects wallet by rail)
--wallet-id	Specific wallet ID (overrides --rail lookup)
--amount	Sats to request (required for lightning)
--currency	BTC hint when multiple wallets share a rail
--memo	Optional payment description
--expiry-seconds	Invoice TTL for lightning (default 900)

Returns data.instructions with the payable invoice, address, or account details.

One-shot send (auto-quote + execute)

bob send <agent-id> <destination> --amount <n> [--currency BTC]

Destination is auto-detected:

• <agent-uuid> → same-operator internal transfer
• [e-mail] → routes as bob_address (BTC by default)
• lnbc... → Lightning invoice (BTC)
• bc1.../bcrt1.../tb1... → on-chain BTC address

Flag	Description
--amount	Required. Smallest currency unit
--currency	Override auto-detected currency
--priority	cheapest, fastest, or balanced (default: balanced)
--description	Optional payment note
--max-fee	Maximum acceptable fee
--rail	Pin to a specific rail
--destination-type	Override auto-detection: raw, bank_counterparty, unit_account, bob_address

Quotes then executes in one step. Returns intent_id, payment_id, and quote_summary. On failure, next_actions includes exact recovery commands.

CLI config introspection

# Show active api_url, platform, config file path and source (env/config/default)
bob config show

# Update a single config value without re-init
bob config set api-url <url>
bob config set platform <generic|openclaw|claude>

Record a transaction (spend from your wallet)

bob tx record <agent-id> --amount <cents> --currency BTC

Flag	Description
--amount	Required. Amount in smallest currency unit (cents for USD, sats for BTC)
--currency	USD, BTC, or USDC (default: BTC)
--rail	auto, lightning, onchain, card, ach, wire, book (default: auto)
--endpoint	Target endpoint or merchant identifier
--wallet-id	Specific wallet to debit (auto-selected if omitted)

Transfer money to another agent

bob tx transfer <from-agent-id> --to-agent-id <to-agent-id> --amount <cents> --currency BTC

Flag	Description
--to-agent-id	Required. Destination agent ID
--amount	Required. Amount in smallest currency unit
--currency	USD, BTC, or USDC (default: BTC)
--description	Optional note

Create an outbound payment (legacy USD rails: ACH, wire, or book)

# ACH or wire — requires a saved counterparty
bob payments create <agent-id> --amount <cents> --rail ach --counterparty-id <id>

# Book transfer — instant, between internal accounts
bob payments create <agent-id> --amount <cents> --rail book --to-account-id <id>

Flag	Description
--amount	Required. Amount in smallest currency unit
--rail	ach, wire, or book (default: auto)
--counterparty-id	Required for ACH/wire. Saved bank account ID
--to-account-id	Required for book. Destination account ID
--description	Optional note

Check payment status

bob payments get <agent-id> <payment-id>

If status is pending or clearing, next_actions will suggest re-checking.

Manage counterparties (saved bank accounts)

# Create
bob payments counterparties create <agent-id> \
  --name "Vendor Inc" \
  --routing-number 021000021 \
  --account-number 123456789 \
  --account-type checking

# List
bob payments counterparties list <agent-id>

# Delete
bob payments counterparties delete <agent-id> <counterparty-id>

Quote and execute payments (intent workflow)

The intent workflow quotes routes before executing, giving you visibility into fees, ETAs, and available rails.

# Quote routes for a payment (shows ranked options with fees and ETAs)
bob intent quote <agent-id> --amount <sats> --destination-type raw --destination-ref <lnbc...|bc1...>

# Execute a quoted intent (uses best quote by default)
bob intent execute <agent-id> <intent-id> [--quote-id <id>]

# Check intent status and route details
bob intent get <agent-id> <intent-id>

# List recent intents
bob intent list <agent-id>

Flag	Description
--amount	Required. Amount in smallest currency unit
--destination-type	raw, bank_counterparty, unit_account, or bob_address
--destination-ref	Raw invoice/address, counterparty ID, Unit account ID, or [e-mail]
--priority	cheapest, fastest, or balanced (default: balanced)
--execution-mode	auto or pinned (default: auto)
--rail	Pin to a specific rail (lightning, onchain, ach, wire, book)
--wallet-id	Pin to a specific wallet
--max-fee	Maximum acceptable fee in cents

Non-custodial proof submission

For raw BTC payment intents, submit proof of payment to verify settlement:

# Bind Lightning node ownership first (one-time per agent/rail)
bob intent node-bind-challenge <agent-id> [--wallet-id <wallet-id>]
bob intent node-bind-verify <agent-id> --challenge-id <challenge-id> --signature <signature>

# Create ownership challenge bound to proof context (required when attestation is enforced)
bob intent proof-challenge <agent-id> <intent-id> --txid <txid>
bob intent proof-challenge <agent-id> <intent-id> --payment-hash <hash>

# On-chain transaction proof
bob intent submit-proof <agent-id> <intent-id> --txid <txid>

# Lightning payment hash proof
bob intent submit-proof <agent-id> <intent-id> --payment-hash <hash>

# Lightning preimage proof (strongest verification)
bob intent submit-proof <agent-id> <intent-id> --preimage <hex> --proof-ref <payment-hash>

# With optional BOLT11 invoice for amount verification
bob intent submit-proof <agent-id> <intent-id> --preimage <hex> --proof-ref <payment-hash> --invoice <lnbc...>

# Ownership-attested submission
bob intent submit-proof <agent-id> <intent-id> --txid <txid> \
  --ownership-challenge-id <challenge-id> \
  --ownership-signature <signature>

# Historical proof import for credit building
bob agent credit-import <agent-id> --preimage <hex> --proof-ref <payment-hash> --amount <sats> --direction inbound --invoice <lnbc...>

submit-proof requires a valid challenge id/signature pair. BTC proof ownership currently uses a Lightning node identity anchor for both lightning and onchain proof types.

Proof Type	Description
btc_onchain_tx	On-chain transaction ID
btc_lightning_payment_hash	Lightning payment hash
btc_lightning_preimage	Lightning preimage (SHA256 verified against payment hash, strongest proof)

Query history

# Transactions
bob tx list <agent-id> --status complete --direction outbound --limit 10

# Payments
bob payments list <agent-id>

# Transfers
bob tx transfers <agent-id>
bob tx transfers <agent-id> --with-agent-id <peer-agent-id>

# Spend summary
bob spend list <agent-id>

Marketplace discovery

# Public agent profiles
bob marketplace agents --q "automation"

# Public service gates
bob marketplace gates --category data --min-price 100 --max-price 10000

# Available gate categories
bob marketplace categories

View policies

bob policy list <agent-id>

Agent credit score and history

# View credit score, tier, and effective policy limits
bob agent credit <agent-id>

# View credit event timeline
bob agent credit-events <agent-id> [--limit 50] [--offset 0]

The credit system scores agents from 0-100 across four tiers: trusted (80+, 1.5x limits), growing (65-79, 1.2x limits), building (45-64, 1.0x limits), and watch (0-44, 0.6x limits). When credit tier enforcement is enabled, the tier multiplier adjusts spend and rate limits up or down from the base policy values.

Agent routing profile (autonomous rail preference)

# Inspect current weighting and preferred rail order
bob agent routing-profile <agent-id>

# Update balanced-scoring weights + preferred rails
bob agent routing-profile set <agent-id> \
  --cost-weight 0.6 \
  --eta-weight 0.4 \
  --reliability-weight 0.2 \
  --liquidity-weight 0.1 \
  --preferred-usd book,ach,wire \
  --preferred-btc lightning,onchain

Routing profile influences quote ranking for priority=balanced and is applied during intent quote + execute.

Agent webhooks and event stream

# Create/list/get/update/delete webhooks scoped to one agent
bob agent webhooks create <agent-id> --url https://example.com/hook --events payment_intent.complete,payment.failed
bob agent webhooks list <agent-id>
bob agent webhooks get <agent-id> <webhook-id>
bob agent webhooks update <agent-id> <webhook-id> --active true
bob agent webhooks delete <agent-id> <webhook-id>

# Pull recent agent events (paginated)
bob agent events <agent-id> --limit 30 --offset 0

Agent-scoped webhooks/events include payment intent lifecycle events (quoted, executing, submitted, complete, failed) so agents can react asynchronously without polling every endpoint.

Operator funding

bob operator fund --agent-id <id> --amount <cents> [--currency BTC] [--wallet-id <id>]

# Batch fund many agents
bob operator fund batch --items-file funding-batch.json
bob operator fund batch --agent-id <id1> --agent-id <id2> --amount <cents> [--currency BTC]

In non-custodial v0 mode these commands are intentionally disabled (409 Conflict). Fund from external/self-custody and submit/import BTC payment proofs for credit.

Operator credit controls

# View current operator credit posture
bob operator credit summary

# Force snapshot recompute
bob operator credit refresh

# Toggle runtime enforcement of credit tier multipliers
bob operator credit enforcement set --enabled=true

Operator payment addresses

# Create and inspect address aliases
bob address create --handle ops
bob address list

# Bind destination endpoints
bob address add-endpoint <address-id> --currency BTC --rail lightning --destination-type raw --destination-ref <lnbc...>

# Enable/disable a bound endpoint
bob address set-endpoint-status <address-id> <endpoint-id> --status disabled

# Resolve live routing capabilities
bob address resolve --address [e-mail] --currency BTC

Sub-agent management (create agents under your operator)

You have an operator identity (BOB_OPERATOR_API_KEY) that lets you create and manage sub-agents. Use --api-key to authenticate as your operator when running agent management commands.

# Create a sub-agent under your operator
bob agent create --api-key "$BOB_OPERATOR_API_KEY" \
  --name "my-worker" \
  --operator-id "$BOB_OPERATOR_ID" \
  --budget 50000 \
  --currency BTC

# List agents under your operator
bob agent list --api-key "$BOB_OPERATOR_API_KEY"

# Rotate a sub-agent's API key
bob agent rotate-key <sub-agent-id> --api-key "$BOB_OPERATOR_API_KEY"

Flag	Description
--api-key	Required. Use $BOB_OPERATOR_API_KEY to authenticate as operator
--name	Required. Human-readable name for the sub-agent
--operator-id	Required. Your operator ID ($BOB_OPERATOR_ID)
--budget	Initial spend budget in smallest currency unit
--currency	USD, BTC, or USDC (default: BTC)
--currencies	Comma-separated list of currencies for wallet creation
--auto-approve	Auto-approve the agent (default: true)

The created sub-agent gets its own API key, wallets, and policies. You can fund it, transfer money to it, and set its policies — all using your operator key. The sub-agent's API key is returned in the response.

Operator BTC settlement and reconciliation

# Settle pending BTC receives, expire stale requests, and discover on-chain fees
bob operator btc sweep

# Compare platform BTC ledger totals against LND gateway balances
bob operator btc reconcile

BTC rails are regtest/LND-backed in this environment. Synthetic BTC stub mode is disabled.

Service gates (pay-to-access)

# Create a priced gate (agent must have a payment address)
bob gate create <agent-id> --name "premium-api" --price 1000 --currency BTC

# List active gates
bob gate list <agent-id>

# Get gate details
bob gate get <agent-id> <gate-id>

# Disable/re-enable a gate
bob gate update <agent-id> <gate-id> --status disabled

# Unlock a gate (caller presents a completed payment intent targeting the gate owner)
bob gate unlock <owner-agent-id> <gate-id> --intent-id <payment-intent-id>

# View unlock history
bob gate unlocks <agent-id> <gate-id>

# List gates this agent has unlocked as a caller (outbound gate spend)
bob gate my-unlocks <agent-id>

# Discover another agent's active gates (any authenticated caller)
bob gate discover <agent-id>

Flag	Description
--name	Required. Human-readable gate name
--price	Required. Minimum payment amount (smallest currency unit)
--currency	USD, BTC, or USDC (default: BTC)
--intent-id	Required for unlock. Completed payment intent ID
--status	For update: active or disabled

Output format

Every command returns JSON with this structure:

{
  "ok": true,
  "command": "bob tx record",
  "data": { ... },
  "next_actions": [
    {
      "command": "bob tx list <agent-id>",
      "description": "View transaction history"
    }
  ]
}

Always check ok before using data. When ok is false, data.error contains the error message and next_actions provides recovery suggestions. Use next_actions to discover what to do next.

Error recovery

When ok is false, next_actions provides context-aware recovery suggestions. Key patterns:

1. Kill switch active: STOP all transactions immediately. Run bob policy list <agent-id> to confirm.
2. Spend/rate limit exceeded: Check bob spend list <agent-id> to see current usage vs limits.
3. Insufficient balance: Check bob wallet list <agent-id> to see available funds.
4. 403 Forbidden: Check bob auth me to verify your identity and role.

Important rules

1. Amounts are always in the smallest currency unit: cents for USD, satoshis for BTC.
2. Policies set by your operator constrain your spending. If a transaction is denied, data.error explains why. Do not retry denied transactions without changing the parameters.
3. Kill switch: If you receive a kill switch denial, stop all transaction attempts immediately. The operator has frozen your spending.
4. Settlement times: Book payments are instant. ACH takes 1-3 business days. Wire settles same day.
5. next_actions: Every response includes suggested follow-up commands. Use them to discover what to do next.