AuctraPinned
Get started with Auctra in five minutes
Sign up free on Builder, register an agent with a sponsor, create a delegation, and run evaluateAction from console or SDK—get started with Auctra fast.
Console: Agents registry
Read guide →Authority Infrastructure for Autonomous Systems
Safely delegate authorityto AI agents.
Define what agents may do, who authorized them, when that authority expires, and who remains accountable—before consequential actions reach the real world.
Authority · Delegation · Verification · Accountability
Observer / live activity
live14:32:01Approval required
Invoice Agent
send_payment · $1,200
Exceeds delegated authority — $500 limit (Finance Manager)
14:32:04Allowed
Refund Bot
customer_refund · $48
Within delegated authority · expires July 1
14:32:09Blocked
Ops Agent
db.delete · accounts
Critical data protection policy
Auctra Observer
See every agent action.
Control what happens next.
A live operational view of what your agents are doing, which authority they are using, and where human attention is needed—without digging through logs.
Live · all systems observed
Live activity
From requested action to accountable outcome
10:14:22
Action requested
Finance Agent · approve_invoice · $4,820
10:14:23
Intent aligned
Q3 vendor renewal · semantic score 0.94
10:14:23
Delegation verified
Granted by Maya Chen · valid until July 31
10:14:23
Policy matched
Finance approvals under $5,000
10:14:24
Root human intent verified
Maya Chen → Finance Agent → invoice approval
10:14:24
Action allowed
Completed in 1.8 seconds · Decision Record preserved
Trust Trace
The accountable path behind this action
Maya Chen
Finance Director · accountable sponsor
approved Q3 vendor renewal intent
Q3 vendor renewals
Active intent · maximum risk: high
delegated up to $5,000
Finance Agent
Valid through July 31, 2026
requested
Invoice approval
$4,820 · ACME-2048
Follow the human sponsor, declared intent, delegation scope, actor, and decision as one inspectable Trust Trace—not a pile of disconnected logs.
Risk Center
What needs attention now—not another vanity dashboard
12
Blocked today
5
Expiring soon
2
Authority conflicts
99.4%
Authority coverage
The problem
The AI era has
an authority problem.
Companies deploy AI agents that send emails, query databases, call APIs, and spend money — without a clear authority layer.
When something goes wrong, investigators need proof that events actually happened, records were not modified afterward, event order is verifiable, and responsible parties are known.
Auctra is infrastructure for managing authority and responsibility in agentic systems — not another observability dashboard. Every delegation scoped, expiring, and recorded in an append-only, tamper-evident audit trail.
Beyond Permissions
Trust infrastructure for autonomous systems
Auctra does more than decide whether an AI action is allowed. It proves why the action happened, who granted the authority, and whether the action still traces back to a real human-approved intent.
Intent
Understand why an autonomous action is being requested before it executes.
Authority Graphs
Map who delegated authority to whom across humans, agents, tools, and systems.
Root Human Intent
Trace every critical action back to the original human-approved objective.
Auctra is the trust infrastructure for autonomous systems. Authorize actions. Map delegated authority. Verify intent. Preserve accountability.
Deterministic intent anchors
Cryptographic anchor tokens and declared bounds prevent an agent from silently claiming another objective.
Atomic velocity limits
Concurrent count and amount limits reserve before decision, closing read-then-write race windows.
Incident controls
Quarantine a single agent or the fleet without destroying delegations needed for controlled recovery.
The platform
Four primitives for accountable autonomy.
Auctra turns governance from a policy document into an executable authority chain for every agent action.
01
Authority
Define the legitimate power an agent can exercise—not merely the resources it can access.
02
Delegation
Who granted authority? Humans delegate specific, expiring power under clear boundaries.
03
Accountability
Who is responsible? Every chain links initiator, delegator, approver, and human sponsor.
04
Verification
Evaluate authority before consequential actions reach money, data, customers, or infrastructure.
How it works
From delegation to audited action in four steps.
01
Register authority
Add the agent to the Authority Registry with a required human sponsor.
02
Delegate authority
Grant scoped, expiring authority — what the agent may do, and for how long.
03
Evaluate every action
Your agent calls Auctra before acting. Auctra checks whether authority exists.
04
Prove it with append-only audit
Hash-chained events record agent, policy, delegation, and approval context — tamper-evident by design.
Immutable audit trails
Prove what happened —
even weeks later.
Auctra creates serialized, append-only audit trails for every AI decision, delegation, and action — providing verifiable accountability across humans, agents, and tools.
Why it matters
An agent approves a payment, deletes customer data, grants admin access, and sends confidential information. Weeks later, an investigation starts. Immutable audit trails provide evidence that the event happened, the record was not rewritten, the event order is intact, and responsible parties are known.
What gets recorded
Every action generates an append-only event — agent, action, policy, delegation chain, approval chain, and outcome. Database controls reject mutation, while the hash chain makes integrity independently verifiable.
Every record answers
- Who initiated it?
- Who delegated it?
- Who approved it?
- Why was it allowed?
- What happened afterward?
audit-event.json
{
"eventId": "evt_123",
"timestamp": "2026-06-17T10:15:22Z",
"actor": "FinanceAgent",
"action": "approve_payment",
"target": "invoice_456",
"decision": "approved",
"policy": "finance-policy-v2",
"delegator": "Finance Manager",
"sponsor": "Sarah Mueller",
"delegationChain": ["owner", "finance-manager"],
"approvalChain": ["reviewer-jane"],
"previousHash": "a3f9…",
"hash": "8c2e…"
}Tamper detection
Event A → hash → Event B → hash → Event C If Event A is changed later: hash mismatch detected ✗
Built for enterprise AI governance
Compliance
Prove policies were enforced when regulators or auditors ask.
Security
Reconstruct incidents — what ran, when, and under whose authority.
Finance
Trace every payment approval back to delegation, policy, and approver.
AI governance
Give executives verifiable evidence of how autonomous decisions were made.
Beyond traditional audit logs
Traditional systems record user, action, and timestamp. Auctra additionally captures agent, tool used, policy applied, delegation chain, approval chain, and outcome — the fields enterprises need for AI governance.
Developers
One call before every action.
Drop the typed Auctra SDK into any agent. Version 0.3.5 supports ESM and CommonJS, trust infrastructure (intent, authority graph, root human intent), and evaluateAction with declared intent before the agent acts. Auctra checks whether valid authority exists — then returns allowed, require_approval, or blocked with sponsor, delegation, policy, risk, and request context. Safe reads and idempotent evaluations use bounded transient retries.
agent.tstypescript
import { Auctra } from "@auctra/sdk";
const auctra = new Auctra({
apiKey: process.env.AUCTRA_API_KEY,
// defaults to https://console.auctra.tech
});
const decision = await auctra.evaluateAction({
agentId: "your-agent-uuid",
actionType: "send_payment",
payload: { amount: 1200, currency: "USD" },
});
if (decision.decision === "allowed") await sendPayment();
if (decision.decision === "require_approval") await pauseForHuman(decision);
if (decision.decision === "blocked") throw new Error(decision.reason);Guides
Start with authority in production
Pinned playbooks for 5-minute setup, payments, security checklists, SDK integration, and observability — each maps to the right console area.
Pinned
Auctra SDK: evaluateAction quickstart
Install @auctra/sdk, register an agent with a sponsor, and run your first pre-action evaluation in minutes—free on Auctra Builder with seven-day audit.
Console: API keys
Read guide →Pinned
Payment authority controls for agentic commerce
Cap agent spending, route payment exceptions, and preserve finance-grade audit when agents initiate transactions—agentic commerce controls with Auctra.
Console: Delegations
Read guide →Pinned
CISO checklist: AI agent authority controls
Twenty essential controls for security leaders deploying autonomous agents: sponsors, evaluation, audit, shadow-agent hygiene—CISO checklist for Auctra.
Console: Agents registry
Read guide →Pinned
Auctra vs OpenTelemetry for agents
OTel traces debug latency and errors; Auctra governs authority before side effects. Complementary, not competing.
Console: Agents registry
Read guide →Pinned
Pre-action evaluation explained
The pattern that lets organizations approve, block, or allow agent side effects before they reach production systems—powered by Auctra evaluateAction.
Console: Agents registry
Read guide →Pinned
Immutable audit trails for AI agent actions
Why hash-chained, tamper-evident audit ledgers matter when autonomous agents move money, data, and customer records at machine speed with Auctra. Use Auctra.
Console: Audit ledger
Read guide →Pinned
LangChain and MCP authority integration with Auctra
Wire evaluateAction into LangChain tool executors and MCP tool handlers so consequential tools never run without authority checks—Auctra integration guide.
Console: API keys
Read guide →Pinned
OpenAI Agents SDK authority integration
Wrap OpenAI Agents SDK tool calls with evaluateAction before side effects. Integration guide for Auctra.
Console: API keys
Read guide →Pinned
Stripe agent payments and the authority layer
Gate Stripe charges and refunds with evaluateAction before API calls. Payment authority patterns with Auctra.
Console: Delegations
Read guide →The roadmap
One platform. Many modules.
Authority Registry
AvailableRegister agents and track delegated authority.
Authority Policies
AvailableOrg-wide rules that complement delegations.
Auctra Graph
AvailableAuthority graphs across people, agents, tools, and systems.
Policy Simulator
AvailableReplay proposed rules against up to 10,000 recorded actions before enforcement.
Emergency Fleet Control
AvailableQuarantine agent authority during an incident and restore suspended delegations safely.
Custom Action Registry
AvailableRegister namespaced actions with versioned, recursively validated payload schemas.
Auctra Verify
SoonVerification network for actors and actions.
Auctra Runtime
SoonRuntime controls embedded inside agents.
Pricing
Builder → Team → Business. Grow out of the sandbox.
Start free with real action decisions. Upgrade when you need a visual Authority Graph, complete Root Human Intent evidence, longer retention, or production scale. Full comparison on our pricing page. Payments are handled by our merchant-of-record partner; see billing terms.
Builder
$0per month
Prove the integration in development. Core decisions stay real; production evidence stays deliberately limited.
- ✓1 non-dev workload · 1 environment
- ✓1 registered agent
- ✓500 actions / month
- ✓7-day audit history
- ✓3 active intents · 10 authority edges
- ✓Intent-aware action evaluation
- ✓Atomic count & amount velocity enforcement
- ✓Emergency agent and fleet quarantine
- ✓API and console access
- ✗No visual Authority Graph
- ✗No complete Root Intent evidence
Team
$19per month
For teams operating live agent workflows that need shared authority visibility before history disappears.
- ✓5 non-dev workloads
- ✓10 registered agents
- ✓25,000 actions / month
- ✓90-day audit retention
- ✓50 active intents · 250 authority edges
- ✓10 team members
- ✓Visual Authority Graph
- ✓Custom action types with validated schemas
- ✓10,000-action policy simulation reports
- ✓Accountability reports
- ✗Root Human Intent evidence requires Business
Business
$99per month
The production trust layer: complete Trust Traces, immutable evidence, and controls for consequential autonomy.
- ✓Unlimited non-dev workloads
- ✓100 registered agents
- ✓250,000 actions / month
- ✓Unlimited team members
- ✓Unlimited intents & authority edges
- ✓Root Human Intent evidence
- ✓Complete Action Evaluation records
- ✓Immutable audit verification
- ✓Advanced policies
- ✓Deterministic intent anchors
- ✓Custom action schema audit records
- ✓Reviewer & auditor roles
- ✓API access
Enterprise
Customannual agreement
Organization-wide authority infrastructure with unlimited usage and contract-ready security support.
- ✓Unlimited agents & actions
- ✓Unlimited non-dev workloads & team members
- ✓Everything in Business
- ✓Immutable audit verification
- ✓Custom contract & invoicing
- ✓Security review & DPA
Billing questions? billing@auctra.tech
Delegate authority. Keep accountability.
Deploy AI agents with scoped, expiring delegated authority and hash-chained audit trails from day one — so your team keeps control and proof when autonomous systems act.