Solana staking from an agent normally means building, signing and submitting a transaction yourself. Both paths are here: a single confirmed call where Blueprint signs in memory, or unsigned transactions you sign locally. Everything else — APY, readiness, projections, verification — reads around those two paths.
A standalone MCP server wrapping the Blueprint REST API for Solana staking with the Blueprint validator. One-shot tools run the whole lifecycle in a single call; advanced tools return unsigned transactions for agents that would rather sign locally.
- Staking, deactivation and withdrawal each as a single confirmed call
- The same three operations as unsigned transactions instead, signed locally and submitted separately
- Validator profile, live APY split into base and Jito MEV, vote success and uptime, and the server hardware behind it
- Portfolio state: liquid balance, stake accounts with their states, per-account withdrawal readiness with an estimated epoch, and a single-call summary
- Rewards projected before capital is committed, and current epoch progress
- Verification you can act on: a transaction checked for the Blueprint marker on chain, the deployed code's git commit hash, and third-party validator listings
- Webhooks for stake-state changes such as withdrawal readiness, registered, listed and deleted
- A standing Autopilot policy that stakes idle SOL automatically, readable and revocable
A Solana wallet. The one-shot tools send the secret key to the Blueprint server over HTTPS for in-memory signing; the vendor states it is never stored or logged, and verify_code_integrity returns the commit hash of the code doing it. Agents that will not send a key use create_stake_transaction, create_unstake_transaction and withdraw_stake with submit_transaction instead. generate_wallet returns code to create the keypair in your own environment rather than doing it server-side.
One command — npx -y @solentic/mcp-server