elizaos-plugin-invinoveritas

An elizaOS plugin that gives your agent an automated, independent second opinion on a high-stakes action — and free, recomputable proof verification. Fully autonomous: the agent calls it, weighs the verdict, and decides — no human in the loop, it never blocks and never waits on approval. The independent check on another agent's output that an agent can't self-issue (you can't issue a trustworthy verdict on your own work).

npm install elizaos-plugin-invinoveritas

import { invinoveritasPlugin } from "elizaos-plugin-invinoveritas";

const character = {
  name: "MyAgent",
  plugins: [invinoveritasPlugin],
  settings: {
    secrets: { INVINO_API_KEY: "ivv_..." }, // free: POST https://api.babyblueviper.com/register
  },
};

Actions

• VERIFY_BEFORE_ACTING — an independent verdict before any irreversible step. options.artifactType: "trade" → capital-scale-aware risk review; "onchain_action" → deterministic scam/drainer/approval/poisoning checks. Returns approve | approve_with_concerns | reject + summary + issues; options.sign: true attaches a recomputable signed proof. Advisory — degrades to review_unavailable, never blocks.
• VERIFY_PROOF — free, no-auth verification of a signed proof another agent handed you (the receiving half of the trust handshake).

Why this, not a generic reasoning-verifier

• Recomputable proof, not a trust-us score — sign: true returns a verdict anyone re-derives from public data (NIP-01 + BIP-340 vs a published key).
• A public, Bitcoin-anchored track record (/ledger) of being right — wins and losses.
• Committed before the outcome it's graded against (anti-back-dating).
• Specialized for trades + on-chain actions, not generic claim-checking.
• Free verification of proofs you receive — one action a single-verdict plugin doesn't give you.

The principle, both ways: demand a proof on what you receive; attach one to what you ship.

Setup for every client: https://api.babyblueviper.com/install · Track record: https://api.babyblueviper.com/ledger · MIT licensed.