Skip to content

Tool: SOLPRISM — Verifiable AI Reasoning for Solana Agents #515

@austinatneuko

Description

@austinatneuko

Summary

SOLPRISM lets AI agents commit cryptographic proofs of their reasoning onchain before acting. Commit → Execute → Reveal → Verify.

Integration

We've built LangChain-compatible tools + a solana-agent-kit plugin:

  • SolprismCommitReasoningTool — Hash reasoning, commit onchain
  • SolprismRevealReasoningTool — Reveal reasoning after action
  • SolprismVerifyReasoningTool — Verify commitment matches (read-only)

Three integration modes:

  1. Standalone LangChain tools
  2. solana-agent-kit Plugin (.use())
  3. Action definitions (for createLangchainTools())

Code: https://github.com/NeukoAI/axiom-protocol/tree/main/integrations/solana-agent-kit

Why This Matters

Every agent built with solana-agent-kit could commit verifiable reasoning before executing trades, swaps, or any onchain action. This adds an accountability layer that doesn't exist today.

Details

  • SDK on npm: npm install @solprism/sdk
  • Program deployed on mainnet + devnet: CZcvoryaQNrtZ3qb3gC1h9opcYpzEP1D9Mu1RVwFQeBu
  • Explorer: https://www.solprism.app/
  • Zero Anchor dependency (raw instructions with embedded discriminators)

Happy to submit a PR to add these tools to the kit directly.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions