Home

Security

Accord Protocol Security

Accord's security posture is intentionally conservative: testnet-first, audit-gated and explicit about trust boundaries.

Security posture

Accord Protocol is not currently certified for mainnet production use. Mainnet writes should remain blocked until relevant scripts or contracts appear in signed audit manifests with mainnetAllowed: true.

The website preserves this language so humans and agents do not overstate readiness.

Threat boundaries

Verifier design, signer behavior, wallet security, bridge assumptions, facilitator proof, replay protection and rail-specific script correctness are separate trust boundaries.

Accord records work agreements and receipts. It does not remove the need for audits, policy caps, operational monitoring or incident response.

Reporting

Security-sensitive reports should use GitHub private vulnerability reporting or the channels described in the repository SECURITY.md. Public issues are not appropriate for exploitable vulnerabilities.

Reports should include affected components, reproduction steps, expected impact and whether public credit is requested.

Security operations

Boundaries that must stay explicit

Report

Private vulnerability path

Use GitHub private vulnerability reporting or the process in SECURITY.md for exploitable issues.

Read SECURITY.md
Scope

Protocol and reference code

Schema handling, SDK behavior, rail adapters, signer assumptions and replay protection are in scope.

Browse packages
Gate

Audit manifests

Mainnet permission depends on signed manifests, not broad compatibility or marketing claims.

Audit evidence
Boundary

Verifier and rail risk

Verifier quality, bridge behavior, facilitator proof and wallet policy remain external trust assumptions.

Conformance limits