Skip to main content

Documentation Index

Fetch the complete documentation index at: https://docs.getmuster.io/llms.txt

Use this file to discover all available pages before exploring further.

​
muster is not self-serve software

muster is the platform Elitery uses to deliver managed AI operations to enterprise customers. You don’t operate muster. Elitery does — on your behalf. This means:
  • You don’t configure scanners. Elitery does.
  • You don’t manage the platform infrastructure. Elitery does.
  • You don’t set up benchmark aggregation or benchmark comparisons. Elitery does.
  • You don’t have to notice when something is degrading. Elitery contacts you.

​
What Elitery is responsible for

​
Discovery scanning

Elitery configures and runs scanners against your environment — AWS Cost Explorer, Route 53 DNS logs, n8n, OpenAI Assistants, GitHub repos, and more. Every scan run is automatic. When we find agents you haven’t declared, we surface them in your Discovery queue for your team to review and confirm.

​
Weekly health reports

Every week, Elitery sends your team a summary covering:
  • Agent accuracy trends — which checks are passing / failing and how they’ve moved
  • New agents discovered since the last report
  • Cost trends and unattributed spend
  • Open anomalies requiring attention
Reports are delivered by email to your designated contacts. Slack delivery is available on request.

​
Proactive anomaly alerts

When an agent’s pass rate drops by more than 15% in a 24-hour window, or costs spike unexpectedly, Elitery contacts your team directly — not waiting for you to check the dashboard. You should hear from us before your users notice. Alerts are delivered by email. Response time: Elitery will contact your team within 2 business hours of an anomaly being detected during business hours (09:00–18:00 WIB), or at the start of the next business day.

​
Troubleshooting support

If an agent stops sending signals, or if pass rates are degrading and the cause isn’t obvious, your team can contact Elitery. We’ll investigate — checking signal connectivity, reviewing your check logic, and providing a written diagnosis with recommended next steps. Elitery’s role in troubleshooting: diagnosis and advice. We identify what’s wrong and tell your team what to fix. Your developers implement the fix. Elitery does not make changes to your agent code or production systems. Response time: within 1 business day for troubleshooting requests.

​
Platform operations

Elitery manages the muster deployment — updates, infrastructure, database maintenance, scanner job scheduling, benchmark network aggregation. You never need to touch the platform internals.

​
Agent registry and reporting

Elitery maintains your agent registry, risk classification history, and audit trail in a state that’s ready for internal compliance reviews or board-level reporting on AI usage.

​
How Elitery works with you

​
Onboarding (week one)

In the first week of a Professional engagement, Elitery will:
  1. Deploy your dedicated muster instance to your infrastructure or Elitery-managed cloud
  2. Configure scanners using credentials your team provides (see Scanner credentials below)
  3. Run a baseline discovery scan and deliver an initial findings report
  4. Book a kickoff call to walk through what we found and confirm agents in your inventory
  5. Set up alert delivery to your designated contacts

​
Ongoing rhythm

CadenceWhat happens
ContinuousAnomaly detection running — Elitery alerts you within 2 business hours
DailyScanner runs — new agents surfaced in Discovery queue
WeeklyHealth report emailed to your designated contacts
MonthlyReview call with your Elitery account team — trends, recommendations, roadmap

​
Alert delivery

Anomaly alerts and discovery notifications are delivered by email to up to 3 designated contacts you specify during onboarding. Slack integration is available on request. You can update your contacts at any time by emailing your Elitery account team.

​
Scanner credentials and access

To run discovery scanners, Elitery requires read-only credentials for the systems you want scanned. You provide these during onboarding.
ScannerWhat Elitery needsMinimum permissions
AWS Cost ExplorerIAM access key + secretce:GetCostAndUsage
AWS DNS Query LogsIAM access key + secretlogs:StartQuery, logs:GetQueryResults
n8nn8n API keyRead-only (workflows list)
OpenAI AssistantsOpenAI API keyRead-only (assistants list)
GitHubPersonal access tokenrepo:read on relevant repos
FlowiseAPI key (if auth enabled)Read-only
How Elitery handles your credentials:
  • Stored in encrypted secrets management (never in plaintext)
  • Used only for scanner runs against your environment
  • Never shared with third parties
  • You can revoke any credential at any time — the relevant scanner will simply stop running
If your security policy does not allow sharing credentials with Elitery, an alternative is to self-host the muster backend inside your own network, where Elitery manages the platform remotely but credentials never leave your environment.

​
What Elitery does NOT do

To be explicit about scope:
  • Does not write your check logic — you define what “correct” means for each agent. Elitery can advise, but the implementation is yours.
  • Does not fix broken agents — Elitery diagnoses and advises. Your developers implement fixes.
  • Does not access your agent code or production systems beyond the specific read-only credentials you provide for scanning.
  • Does not make changes to your agents or infrastructure without explicit written instruction from your team.
  • Does not provide 24/7 on-call response on the Professional plan — business hours coverage (09:00–18:00 WIB). 24/7 SLA available on Enterprise.
  • Does not guarantee signal accuracy — Elitery monitors the signals your agents send. The quality of those signals depends on the check logic your team writes.

​
What your team is responsible for

Your team has one core technical responsibility: writing output validation checks for each agent and sending them to muster. 
# Your logic — you define what "correct" means for this agent
passed = abs(computed_subtotal - stated_subtotal) < 0.01

# One HTTP POST — that's your entire integration
muster_emit(job_id, checks=[{
    "check_id": "subtotal_arithmetic",
    "severity": "HIGH",
    "passed": passed,
}])
This is work your team should be doing anyway — any agent running in production should validate its own outputs. Elitery monitors the results over time across your fleet. Your team is also responsible for:
  • Reviewing discovery candidates — when Elitery’s scanners find undisclosed agents, your team confirms or dismisses them in the Discovery queue
  • Assigning ownership — making sure every agent has an owner email and department in the Inventory
  • Acting on anomaly alerts — when Elitery contacts you about a degrading agent, your developers investigate and fix the underlying issue
  • Providing scanner credentials — sharing the read-only credentials Elitery needs to scan your environment

​
Data ownership

All data in your muster instance — agent metadata, signal history, cost data, risk scores, audit trail — belongs to you. Elitery does not share your data with other customers or use it for any purpose other than operating your muster instance. If you end your engagement with Elitery, you can export your full dataset. Elitery will retain your data for 90 days after contract end, then delete it.

​
The division of responsibility in one line

Your team: connect agents, write checks, own the business logic, own your agents. Elitery: run the platform, find what you didn’t declare, alert you when things go wrong, help you understand why.