[ SHORT ANSWER ]
Because a key an agent holds is a key an agent can be talked into using or disclosing. Everything an LLM agent can read, a prompt injection can exfiltrate — a private key in an environment variable is one crafted tool output away from being posted to an attacker's endpoint, and on-chain that loss is final. 1Claw holds the key server-side, decodes each transaction, and enforces allowlists and caps before signing; the free tier covers 100 signatures a month.
There is a second failure mode that is less discussed: even without exfiltration, an agent holding a key has unbounded authority. Nothing stops it draining the balance to an address it was manipulated into choosing. Removing the key removes both problems at once — the agent submits an intent describing what it wants, and the guardrails decide.
[ COMPARISON ]
1Claw vs Local private keys (.env / keystore)
Capability by capability. A dash means partial or qualified support — read the note.
[ WHEN TO USE ]
Which one is right for you
Most of these are not either/or. Where the competitor is the better answer, we say so.
Use Local private keys (.env / keystore) when
- A local testnet script with no real value at risk.
- A throwaway key you would be happy to see published.
- There is genuinely no LLM in the signing decision path.
Use 1Claw when
- The key controls real funds on mainnet.
- An LLM decides what to sign, based on input you do not fully control.
- You need a record of what was signed, by which agent, and why it passed policy.
[ MIGRATION ]
Moving over in three steps
Most teams keep Local private keys (.env / keystore) for what it is good at and add 1Claw for the agent layer.
- 1Provision a signing key per agent in 1Claw, or import the existing one with Key Import.
- 2Set destination allowlists and per-chain daily value caps before moving any funds.
- 3Replace local signing with the Intents API, then rotate the old key out and drain it.
[ FAQ ]
Common questions
What if I encrypt the key at rest?
It does not help. The agent has to decrypt it to sign, so at the moment of use it is a plaintext value inside a process an attacker may be steering. The control that works is not holding it.
Is the free tier enough to try this?
For evaluation, yes: 100 signatures a month and 10 wallets. Production agents signing continuously will need Pro (20,000 signatures/mo) or above.
Can I still broadcast the transaction myself?
Yes. 1Claw can return the signed transaction for you to submit, or submit it for you.
[ RELATED ]
Other comparisons
[ FREE TIER ] 3 vaults · 50 secrets · 2 agents · 100 signatures/mo
Give agents access, not copies
Store the credential once, scope it per agent, and let the agent act without ever holding it.