[Feature Request] xProof integration - tamper-proof decision anchoring for uAgents #89

Prerequisites

• I checked the existing issues to make sure this feature has not already been requested

Item

Problem

uAgents currently have no mechanism to prove that a decision existed before the action that followed it. Agent logs are self-authored — in any adversarial context (regulatory review, counterparty dispute, insurance claim), self-authored evidence is not independent proof. The agent that acted is also the agent that wrote the record.

Proposal

Add xProof as an optional certification middleware for uAgents. Before each action, the agent anchors the WHY (decision reasoning, confidence level, trigger context) on MultiversX mainnet. After the action, it anchors the WHAT (exact output, result). Both proofs share a decision_id — creating a verifiable confidence trail that no agent can retroactively modify.

Implementation

Already implemented and published:

• Python SDK: pip install xproof[fetchai] (v0.2.4 on PyPI)
• XProofuAgentMiddleware — certify_incoming + certify_outgoing with shared decision_id
• xproof_handler — decorator that wraps any @agent.on_message without touching agent logic
• wrap_agent() — reads agent.name automatically

Reference implementation (runnable, no API key needed):
github.com/jasonxkensei/xproof-examples/tree/main/fetchai

Usage (3 lines)

from xproof.integrations.fetchai import wrap_agent
agent = Agent(name="my-agent")
wrap_agent(agent)

Value

• Any uAgent becomes independently auditable without trusting the agent itself
• Compliance-ready audit trail for regulated verticals (DeFi, trading, insurance)
• 363ms median latency, non-blocking — zero impact on agent execution
• 10 free certifications on registration at xproof.app

References

• xproof.app
• PyPI: xproof[fetchai]
• AGENT_PROOF_STANDARD: github.com/jasonxkensei/xproof-examples

Additional Information (Optional)

No response