How to Configure Flowise SSH Approval

Published
Length
3 min read

Luis Felipe Dias

All posts

You set up flowise ssh approval by placing a human check before shell steps run. The workflow pauses the agent and waits for a reviewer before opening a remote session. Many teams first give an AI agent a static SSH key to a server. That wide role lets a single bad prompt run destructive commands across the host. How flowise ssh approval works in agent workflows Flowise builds agent workflows as graphs of linked nodes. When an agent needs to run a server command, it calls a custom tool. Witho

Approval Chains & Escalation + SSH Access Management: the complete guide

Architecture patterns, implementation strategies, and security best practices. Delivered to your inbox.

Free white paper. No spam. Unsubscribe anytime.

You set up flowise ssh approval by placing a human check before shell steps run. The workflow pauses the agent and waits for a reviewer before opening a remote session. Many teams first give an AI agent a static SSH key to a server. That wide role lets a single bad prompt run destructive commands across the host.

How flowise ssh approval works in agent workflows

Flowise builds agent workflows as graphs of linked nodes. When an agent needs to run a server command, it calls a custom tool. Without a check in place, that tool opens an SSH session right away. To add an approval step, teams break the graph into two parts. The first part lets the language model draft the shell command it wants to run. Then the workflow pauses and saves its current state. It sends the drafted command to a chat channel or ticket queue for review. A human engineer reads the exact string and clicks approve or reject. Only after an approve signal does the graph resume and run the SSH tool.

Teams also pair this workflow check with OpenSSH certificates on the target host. OpenSSH added certificate support for users and hosts in version 5.4. A Certification Authority key signs a public key to make a user certificate. The certificate holds the public key, an identity string, principal names, and options. It uses a much simpler format than X.509 certificates. Unless you set a time window, an OpenSSH certificate stays valid forever. Teams So, sign short time windows right after the human approves the step.

Why host keys and workflow pauses fall short

Native workflow pauses and OpenSSH certificates solve the login part of the problem. They verify who can log in and when the session can start. Still, they leave a major gap open once the session is active. A workflow check only reviews the text inside the graph before the connection opens. If the agent opens an interactive shell or runs a script, host controls go blind. Standard SSH logs show who logged in, not what each command did in flight.

Engineers also struggle when agents read sensitive files over an approved SSH session. A shell output might contain user emails, phone numbers, card numbers, or national IDs. Native SSH tools do not rewrite those matched fields in results before the model reads them. The raw text flows straight back into the agent prompt and external model logs. Also, keeping keys inside workflow variables exposes them to anyone who can edit the graph. Teams need command-level checks and data masking that happen outside the agent app.

Approval Chains & Escalation + SSH Access Management: architecture patterns and best practices

Continue reading? Get the full guide.

Free. No spam. Unsubscribe anytime.

Managing keys and principals for agent hosts

If you rely on native OpenSSH controls, you must lock down the certificate options. When signing a key, the authority must set a clear key identity for the logs. You should also list exact principal names so the agent uses a restricted Linux account. Never let an AI agent share a principal with human admin accounts. By splitting agent principals from human roles, you can trace which actions came from automation. Still, host principals only limit what the Linux user can touch on disk. They cannot pause a single risky command inside an already open SSH session.

Add command and file guardrails to SSH access

Host firewalls and certificate principals control who opens an SSH session. Yet an allowed user or agent can still run a bad one-shot command or pull sensitive files over sftp. In practice, hoop.dev provides an open-source SSH server with guardrails on the host that uses OpenSSH certificates.

Using hoop.dev SSH controls, guardrails check one-shot commands, env variable names, and sftp or scp paths before they run. Port forwards default to deny unless the target is listed. Because interactive shells have no statement boundary and carry no guardrails, teams can turn them off so every command is checked and logged. Check out the MIT-licensed project on GitHub.

Frequently asked questions

Can Flowise pause an SSH tool call on its own?

Yes, you can add a human input step before the custom tool node runs. The graph stops and waits for a user response before it sends the command.

Why are static SSH keys risky for AI agents?

Static keys never expire on their own and grant full shell rights day and night. Anyone who reads the workflow settings can copy the key and log in directly.

Do OpenSSH certificates stop bad commands after login?

A certificate checks the user identity and principal name when the session starts. Once the shell opens, the certificate does not inspect or block individual commands in flight.

Save the open-source gateway.

hoop is MIT-licensed infrastructure for controlling how AI agents reach production data. Star hoophq/hoop so you can inspect it, deploy it, or share it when your team starts governing agent access.