Guide // MandateGuard
How do Agent Passport and spending mandates work together?
By the Ingenious AI team · Published 2026-11-14 · Reviewed 2026-11-14
▸ Run the free live demo See pricing
Free covers 1 agent and 1 policy with the full authorize flow, signed receipts and the approval inbox — no card required to start.
Short answer
Agent Passport answers who an agent is; MandateGuard answers what it may spend. They compose at the agent record: a MandateGuard agent can carry its passport_id (set at creation or via the passport endpoint), and MandateGuard's passport check lets identity standing inform spending operations with a posture you decide in advance for each status. Keep the claims separate — a receipt is not identity proof, a passport is not spend authority — and run identity and mandate reviews in the same sitting, so an identity suspension and a spending freeze happen together rather than in different weeks.
What is the difference between identity and mandate?
An Agent Passport ID identifies which registered agent identity an agent record claims; MandateGuard stores it on the agent (passport_id) and its passport check can consult the identity's status (for example active, suspended or revoked, verified or not) as part of the spending context. The mandate governs what the identified agent may spend. Neither substitutes for the other: identity without a mandate is a named agent with unbounded spend; a mandate without identity is a bounded stranger.
Two questions every agent deployment must answer
Who is this agent? and What may it spend? Deployments that blur the two end up answering neither well: a spending tool that also asserts identity becomes a weak registry, and an identity registry that also waves through payments becomes a weak control. The fleet separates them deliberately. Agent Passport is the Know-Your-Agent registry: identity, standing and verification for agents. MandateGuard is the spending-policy layer: mandates, decisions and receipts. The integration point is deliberately thin — MandateGuard agent records can carry a passport_id, and MandateGuard includes a passport check — so each system stays authoritative for its own question while the answers compose at the agent record.
Attaching identity to the spending record
Attachment is a field, set at agent creation or updated afterwards through the dedicated endpoint, holding the passport ID string (trimmed to length) on the agent record. From then on, every receipt the agent generates belongs to a record that names its claimed identity — the join auditors want when an incident crosses the identity/spending boundary. Operationally, treat the passport ID like other identity metadata: set it from the registry's record, not from the agent's self-description, and update it through the same change discipline as policy edits. An agent record without a passport ID is not broken; it is a deployment that has not adopted identity attachment, which governance may accept for internal low-stakes agents and require for counterparty-facing ones.
What the passport check adds
MandateGuard's passport integration exists so identity standing can inform spending operations: whether the identity behind this agent record is in good standing, suspended, revoked or unverified, and what to do about each state. The design space is deliberately operator-chosen — a deployment might merely record identity status alongside decisions for low-stakes internal agents, and take a stricter posture (refuse to treat unverified or non-active identities as authorised spenders) for agents that face counterparties or larger budgets. The lesson from the integration's own test discipline is that each status needs a decided behaviour in advance — allow, deny, fall back — because identity services, like all services, also have unavailable moments, and an undecided fallback becomes an accidental policy.
Governance: run the two reviews together
Identity and mandate reviews should share a calendar slot. When an agent's identity standing changes — suspension, reinstatement, a verification milestone — the same sitting should ask the spending questions: does this mandate still fit this agent, should it be frozen during the identity event, do its caps reflect its current trust? Conversely, mandate incidents (a freeze for anomalous spending) should trigger an identity look: is this still the agent we registered, on the host we expect? The two systems' evidence joins at exactly these moments. Teams that run the reviews separately discover the joins during incidents, which is the most expensive time to discover anything.
What composition must never blur
Keep the claims straight in every document and demo. A mandate receipt proves a spending decision under a policy; it does not prove identity. A passport proves identity standing; it does not authorise a single purchase. Role keys in the intent service authenticate access to mandate state; they are not passports. This strictness is not pedantry — counterparties build on these artefacts, and an artefact that overclaims corrupts decisions downstream. MandateGuard's own scope statements model the discipline: the intent service explicitly does not issue identity credentials, and the passport relationship is a reference, not an absorption. Compose the systems; never confuse their signatures.
Adoption order
Adopt in the order that matches your risk. Most teams start with mandates, because unbounded spend is the urgent exposure; identity attachment follows when agents face counterparties, regulators or scale. When you attach, do it fleet-wide in one pass — a workspace where half the agents carry passport IDs invites exactly the ambiguity the field exists to remove — and record the adoption in the policy register: which agent classes require passports, what posture applies to each identity status, and who owns the joint review. From there the composition is quiet: identity in the registry, authority in the mandate, evidence in the chain, and an agent record that ties them together by one honest field.
Five actions to take now
- Attach the passport ID when you create or update the agent record
- Decide your posture for unverified agents before enabling stricter checks
- Keep mandate and identity reviews separate — different owners, different evidence
- When identity status changes, review the agent's mandate in the same sitting
- Never present a mandate receipt as identity proof, or a passport as spend authority
Worked example: identity incident, spending response, one sitting
A platform team registers its procurement agent in Agent Passport, obtaining its passport ID, and attaches that ID to the agent's MandateGuard record at creation. The mandate is fitted as usual: caps, allowlist, threshold. Two quarters later the identity review suspends the passport during a credential-rotation incident. The team's runbook treats identity suspension as a spending event: they freeze the agent in MandateGuard in the same sitting, investigate, rotate credentials, and restore both — passport reinstated, agent unfrozen — with the receipts showing a clean, deliberate gap rather than ambiguous activity.
The composition's value shows in the audit: identity evidence answers who this agent is and its standing over time; mandate receipts answer what it was authorised to do and did. Neither artefact has to pretend to be the other.
Frequently asked questions
How do I attach a passport ID?
At agent creation (passport_id in the create call) or later through the agent passport endpoint with workspace credentials. It is stored on the agent record.
What does the passport check do?
MandateGuard's passport integration can check the identity status associated with the agent's passport ID, so identity standing can inform how the deployment treats the agent. Posture choices are the operator's.
What if an agent has no passport ID?
It operates under its mandate as before — the passport attachment is optional. Your governance decides which agent classes require one.
Does a verified passport widen the mandate?
Not automatically. Any cap or allowlist change remains a deliberate policy decision. Identity standing informs governance; it does not silently edit budgets.
Where do I get a passport ID?
From Agent Passport, the fleet's Know-Your-Agent registry — a separate product from MandateGuard, by design.
Keep reading
- How do you set up your first AI agent spending policy?
- How does an AI agent kill-switch work? Freezing spend instantly
- How do role-scoped keys work for issuers, merchants and agents?
Put the policy in force: spend envelopes · intent registry · approval inbox · how the proof works · all MandateGuard guides