Skip to content
Security · Threat ModelOpenAgentID

Threat
model

A practical threat model across cryptographic, identity, delegation, network, and availability domains. Each vector maps to a mitigation strategy implementers can apply and verifiers can enforce.

VerificationDistinct checks. A deliberate boundary.
Categories
5
Vectors
15
Bias
Mitigation-first
§ 01
Radar

Severity plotted from the center out

Center is higher severity. Click a category in the legend to jump to its vectors and mitigations.

§ 02
Vectors

From threat to mitigation

The threats you should expect to face, and the mitigations that keep verification meaningful under adversarial conditions.

severity · CRITICAL

vectors · 3

Key Compromise

Vector

Private key extraction

Mitigation

Prefer hardware-backed storage where available, and keep key material out of logs and diagnostics.

Vector

Memory scraping

Mitigation

Minimize key lifetime in memory; use constant-time comparisons and wipe buffers when practical.

Vector

Side-channel leakage

Mitigation

Use mature implementations and avoid secret-dependent branching in verification paths.

severity · HIGH

vectors · 3

Identity Spoofing

Vector

DID impersonation

Mitigation

Bind identifiers to public keys so impersonation requires the corresponding private key.

Vector

Lineage forgery

Mitigation

Check exact signed parent links using authoritative keys, then current owner/self-root approval and every required revocation. Signature-only offline checks do not establish current trust.

Vector

Document tampering

Mitigation

Canonicalize before signing so any modification invalidates verification.

severity · HIGH

vectors · 3

Delegation Abuse

Vector

Privilege escalation

Mitigation

Design delegation so child authority cannot exceed the parent (least privilege, monotonic narrowing).

Vector

Stale delegations

Mitigation

Prefer time-bounded grants and explicit rotation/revocation paths for long-lived authority.

Vector

Orphaned authority

Mitigation

Define revocation behavior for descendant identities and document the expected verifier behavior.

severity · MEDIUM

vectors · 3

Network Attacks

Vector

MITM on transport

Mitigation

Use authenticated encryption and verifier-side signature checks; never trust the network for authenticity.

Vector

Resolution poisoning

Mitigation

Resolve exact keys and purposes from an approved authority. Schema and a self-signature alone cannot turn a caller-supplied key or owner into trusted identity.

Vector

Replay attacks

Mitigation

Issue audience-bound challenges and atomically consume both nonce and proof ID in durable storage. Short expiry alone does not prevent replay or coordinate replicas.

severity · LOW

vectors · 3

Availability

Vector

Resolution outage

Mitigation

Pinned local material can support signature checks. Required current root, owner and revocation evidence must remain fresh and available; an outage refuses the protected operation.

Vector

Abusive traffic

Mitigation

Rate limit, isolate hotspots, and design systems so verification degrades gracefully under load.

Vector

Dependency failure

Mitigation

Keep required authority and replay stores available. Missing or stale authority fails closed; do not downgrade to syntax, cached approval or signature-only access.

Next

Found an issue in the real world?

Please report vulnerabilities privately so fixes can ship before public discussion.