Skip to main content

Just-in-time access, approved in seconds

ONE PLANE ·SSHKubernetesK8sDatabasesAppsAPIs

Infra access that just works. No security tax. The right access for the right reason. Play back what happened.

See it in action. Try it hands-on. No sales call needed.

almaforge.example.com/resources

Alice finds prod-web-01 in the live resource list. Labels show where each resource runs.

With no standing access, Alice asks for the prod-ssh role for one hour and names the incident.

Alice is on call for the service, so the on-call integration approves her request in seconds.

Alice gets a one-hour certificate for the role and connects to prod-web-01 with ssh.

Bob opens the recordings and plays Alice's session back, exactly as she typed it.

Take the tour →Explore features →

One unified control plane for infrastructure access

Who needs access, to which resource, for how long, and why? Bring identity, on-call context, and review policy into one access decision. Issue short-lived certificates and capture protocol activity. Hosted in your infrastructure.

Engineers↓
AI agents↓
Automation & CI↓
External identity & on-call context
OIDC — Okta, Google, Entra
On-call — PagerDuty
→
AlmaForge unified control plane
Identity
Roles
Reviewers
Resource
Reason
Context-based just-in-time access
→
Continuous governance
Audit logsStructured JSON events
Session recordingsTerminal replay
ephemeral short-lived certificate↓
Servers
prod-web-01
Protected
Kubernetes
eu-cluster-3
Protected
Databases
prod-db-01
Protected
Apps
grafana.internal
Protected
APIs
api.internal
Protected
All protected
Multi-cloud · hybrid · air-gapped · 100% self-hosted · no SaaS · no telemetry

How it works →Security Whitepaper →

Your infra. Your keys. Your rules.

Self-hosted inside your perimeter with zero telemetry, backed by your SSO, and governed by your access policies.

Runs in your network

Self-hosted on-prem, air-gapped, cloud, or hybrid. Your infrastructure. Your data. Your integrations. Not a single byte leaves your environment.

Air-gap readyNo telemetry

One plane, native protocols

SSH, Kubernetes, databases, internal apps, and APIs through one identity-aware gateway. Keep using the tools your team already knows.

Native clientsNo SDK required

Zero Standing Privilege

Grant elevated access within an approved scope and duration. Short-lived certificates expire automatically, keeping production privileges temporary.

Just-in-timeFine-grained

See what happened

Trace access to a person or workload. Replay recorded terminals and send structured protocol events to your SIEM.

Session replaySIEM export

Built to stay up

Multi-node proxies and replicated storage survive zone failures. Automatic failover reroutes traffic to healthy replicas with no single point of failure.

High availabilityRedundancy

Buy once. Keep running.

One perpetual license sized to your deployment. Buy once and keep running your licensed version. Renew for updates and support.

PerpetualNo subscription

Approved in seconds. Closed on its own.

A request names the role, the reason, and the window. Policy decides who has to say yes, and an on-call engineer in an incident does not wait for anyone.

01

On-call approval

When PagerDuty confirms the requester is on call, the request is approved automatically and the grant opens at once.

02

Approval thresholds

Critical roles can require two or more approvals, and a single denial can stop the request.

03

The task in view

Reviewers see the role, reason, and duration on the request. The grant expires on its own.

Watch the session. Follow the evidence.

Replay recorded SSH and Kubernetes terminals, with identity and timing alongside the activity. Database queries and API requests leave structured audit events.

✓

See the real person behind the session

See who connected, even when the target account is shared. Session records identify the person or workload behind the connection.

✓

Audit-ready evidence

Hand the auditor the replay, not a spreadsheet. Recordings stay in your storage, under your retention policy.

✓

Searchable forensics

Search captured queries such as DROP TABLE in your SIEM. Open terminal recordings to review the session.

almaforge.example.com/recordings/7d1e04b2

Bob opens the recordings and plays Alice's session back, exactly as she typed it.

Put AI agents on call. Keep the keys.

Claude Code, Codex, and other agents troubleshoot across Kubernetes and databases through AlmaForge, without compromising security.

✓

Live inventory discovery

Agents list the clusters and databases that exist right now instead of guessing hostnames from stale wikis.

✓

Just-in-time access requests

Agents start with zero standing privilege and request scoped access when a task needs it, for a human or automated approval.

✓

Zero secrets to leak

Short-lived certificates mean no API keys, database passwords, or SSH keys ever enter the model's context.

alice@laptop: ~/checkout
▐▛███▜▌ ▝▜█████▛▘ ▘▘ ▝▝Claude Code v2.1.0 Opus 5.5 · Claude Max ~/checkout
>Checkout requests are timing out since 14:05. Find out why.
⏺Bash(alma kube ls)
⎿Cluster Agent Labels Version ------- ------------------------ ---------------------------- -------- prod-eu 8c1f02d4-5a7e-4b19-9d... env=prod,region=eu-west-1 2026.1.0 staging 2b7e9a10-4c3d-4f8e-a1... env=staging,region=eu-west-1 2026.1.0
✽Thinking… (3s · esc to interrupt)
>
? for shortcuts

Alice asks Claude Code why checkout times out. It starts by listing the clusters AlmaForge lets it reach.

The agent reads the checkout logs on prod-eu. Its database queries time out.

With no standing access to the database, the agent requests read-only access and waits for approval.

The agent finds billing-worker holding a lock that checkout waits on.

The billing-worker logs show why it never commits. The grant expires on its own.

Access as code. Not a security tax.

Controls that feel like friction get routed around through hardcoded secrets, shared keys, and open paths. AlmaForge lives in your Git workflow, so the secure path is the fast path.

Declarative

Roles, policies, and resources are declarative YAML in Git: versioned, reviewed, and rolled back without console clicks.

GitOps-native

Keep your files in Git and apply them with alma apply -f, just like Kubernetes manifests.

Stays in the terminal

ssh, kubectl, psql: native CLI tools with zero portals to babysit.

examples/roles/dev.yamlgit-tracked
apiVersion: almaforge.com/v1
kind: Role
metadata:
name: dev
spec:
allow:
kubernetesLabels:
env: staging
kubernetesGroups: ['developers']
databaseLabels:
env: staging
databaseUsers: ['readonly']
databaseNames: ['orders']
request:
roles: ['prod-*']
thresholds:
- approve: 2

See it in two minutes. Run it in ten.

See it in action. Try it hands-on. No sales call needed. Prefer a guided demo?

Contact us