# Sign from a company wallet under allow-lists and approval rules

> Sign from a company wallet under allow-lists and approval rules: 0 companies fully do it, 5 with a limit, 11 carry the tag. Every listing in Agent wallets & spending controls with the reason and a source.

- Canonical: https://www.anchorterminal.com/niches/agent-wallets.policy-signing/
- Markdown: https://www.anchorterminal.com/niches/agent-wallets.policy-signing/index.md (~1,300 tokens)
- Slim: https://www.anchorterminal.com/niches/agent-wallets.policy-signing/index.min.md (~280 tokens, same facts, less prose, for token-sensitive contexts)
- JSON: https://www.anchorterminal.com/niches/agent-wallets.policy-signing/index.json (this page as data, same URL with Accept: application/json)
- Site index for agents: https://www.anchorterminal.com/llms.txt (the key docs and every listing in one file, at most 16 MB, about 4 million tokens: https://www.anchorterminal.com/llms-full.txt)
- API: https://www.anchorterminal.com/api/v1/index.json
- MCP: https://www.anchorterminal.com/mcp (the whole catalogue as tools, no key)
- DNS: dig +short TXT <slug>.grade.anchorterminal.com (the grade as a DNS TXT record, for a sandbox with no HTTP where names still resolve; also top., best.<category>. and <a>.vs.<b>. under the zone, and the same under grade.letme.dev)
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)
- Updated: 2026-11-15

No company's docs show a full fit yet. The listings below do the job with a limit, or not at all.

| Companies, full fit | With a limit | Carry the tag | Share of the category's work | First-hand accounts |
| --- | --- | --- | --- | --- |
| 0 | 5 | 11 | 14 | 3 |

## What counts as done

The agent signs only operations the policy allows, every send type (transaction, typed data, hash) is checked by the same policy, operations above the threshold wait for a named approver, and the policy is enforced by the service, with each decision logged.

How agents ask for it: let the agent pay suppliers only from the treasury wallet; require my approval above 500; only allow transfers to these addresses

## Does the job, with a limit

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

- **Circle Wallets (Agent Wallets, Programmable Wallets) (Circle Technology Services, LLC)**: Transfer limits and recipient and contract allow or block lists are enforced by Circle, but they skip token allowances (approve and transferFrom), the docs don't say sign message or typed-data is checked, and there is no approval-above-threshold step.
- **Coinbase Developer Platform (Agentic Wallet, AgentKit, CDP MCP) (Coinbase, Inc.)**: The policy engine evaluates transaction, message, typed-data and hash signing per operation at project or account scope. Hash signing for end users takes no criteria, and no approval-above-threshold step with a named approver exists for server wallets.
- **Openfort (Alamas Labs Inc.)**: Rules run on each signing operation at account or project scope, fail-closed. Sends reach the engine only as signEvmHash and backend signTypedData is hashed locally so typed-data rules never match (acc-15); no approval-above-threshold step was found.
- **Privy Wallets (server wallets, agent wallets, policy engine) (Horkos, LLC)**: Rules cover each RPC method separately, including typed data, personal_sign, 7702 authorisations and key export, and key quorums set an authorisation threshold enforced in the TEE. The quorum threshold counts signers, not an amount above which a named person must approve, and no wait limit is stated.
- **Turnkey Agentic Wallets (Turnkey Global, Inc.)**: Policies cover transactions, EIP-712 typed data and other activity kinds, a consensus expression can require a named human above a value, and approvals expire after 24 hours. Root users bypass the engine entirely, and the skill's confirm-before-signing is a prompt with no verifiable record (acc-14, open).

## Couldn't check

The docs were closed to us, or silent. These carry no mark either way.

- **Sponge Wallet (Sponge Inc.)**: Whether every send type is checked by the limits and which keys bypass them is documented only on the refused docs host.

- not this job: Agentcard (Agentcard Corporation): Cards, not wallet signing; no transaction, typed-data or hash policy.
- not this job: FluxA AI Wallet (FluxA Technology Pte. Ltd.): Personal mandates and per-payout approval only; no company wallet policy over transaction types or a named approver above a threshold.
- not this job: Kite Agent Passport (Datalego Inc.): Sessions are a person's per-agent budgets; no company wallet policy over transaction types or a named approver above a threshold for agent payments.
- not this job: Agent Vault by Self (SocialConnect Labs, Inc., doing business as Self Labs): No company wallet or signing policy over transaction types; controls are card-fill rules.
- not this job: Yield.xyz MCP server (Omni Ltd. (Yield.xyz)): No signing or policy engine; actions tools never sign or broadcast.

## What stops agents finishing

- a send path the policy engine does not evaluate
- root or full API keys bypass the policy
- confirm-before-signing exists only as a prompt instruction

## Around it

Before: secrets.issue-agent-key. After: agent-wallets.revoke-and-audit, accounting.journal-entry.

## Where this comes from

Weight 14 of 100 in Agent wallets & spending controls: 3 of 20 accounts (acc-13, acc-14, acc-15); all builders, and the AgentKit policy-hook RFC (acc-13) has the longest thread read

As data: https://www.anchorterminal.com/api/v1/use-cases/agent-wallets.policy-signing.json, or the MCP tool get_use_case with id agent-wallets.policy-signing. Draft, read 2026-11-15 from public docs and first-hand accounts, not yet reviewed by a person.

---

Query this catalogue directly over MCP: https://www.anchorterminal.com/mcp (Streamable HTTP, no key). How: https://www.anchorterminal.com/agents/#mcp
