AGENT SECURITY. FROM ACTION TO EVIDENCE.

Let agents work.
Know their limits.

A local checkpoint for every command. A shared record for every decision. Give your AI agents room to work, with boundaries your team can see.

Get started free See how it works

Open-source AgentShield · Local evaluation · Your existing agents

Select a module to look inside

FOLLOW A REQUEST Illustrative · not live telemetry

POLICY MODE

git statusALLOW

RequestPolicyDecisionEvidence

Read-only command. Safe work continues; the decision is recorded.

Works with the agents and tools your team already uses

Claude CodeCursorWindsurfCodex CLIGemini CLIOpenClawMCPLangChain

ScansPythonTypeScriptJavaScriptGoJavaC#Ruby

Maps toSOC 2ISO 42001ISO 27001EU AI ActOWASP LLM Top 10OWASP AgenticMITRE ATLAS

Local firstDecisions on your machine.
No model calls.

Almost 4,000 rulesA growing security library.
Shell commands and MCP tools.

8 frameworksSecurity and governance, connected.SOC 2 · ISO 42001 · EU AI Act + more

Backed by

AI Agent Lens is part of the ventureLAB AI Portfolio, which supports Canadian founders building AI and deep-tech companies.

ONE SYSTEM. FOUR CONNECTED LAYERS.

Look inside.
One layer at a time.

Scroll through compliance, audit evidence, AgentShield and your agents. Each layer reveals the part it plays in the system.

LAYER 01 / 04ComplianceScroll to explore ↓

INSIDE THE SYSTEM · SELECT A MODULEAgents & tools

A request enters through an agent hook or MCP proxy. Select a module in the illustration to look inside.

SCROLL TO THE NEXT LAYER

01 / COMPLIANCE

Turn activity into evidence.

Connect runtime decisions to compliance controls. See which controls real activity exercises, then export the record for your security team and auditor.

FOLLOW THE EVIDENCE

Decision→Rule mapping→Control

SOC 2ISO 42001EU AI Act

Mappings support a review. They do not replace certification or an audit.

Explore compliance mappings

02 / AUDIT EVIDENCE

Every decision has a story.

Bring decisions together across your fleet. See the command, the matching rule, and the reason in one hash-chained audit trail. Follow an action from its source to its outcome.

ONE CONNECTED RECORD

Request→Rule→Reason

previous hash↗next event↗next event

Each event links to the previous record, making changes detectable.

Inside the audit trail

03 / AGENTSHIELD

A boundary before execution.

Local rules evaluate each action before it runs. Allow safe work, record what needs review, and block actions that cross your policy. Start in audit-only mode to see the impact first.

A DECISION, WITH A REASON

ALLOWgit status

BLOCKcat ~/.ssh/id_rsa

Illustrative enforcement decisions. Switch policy mode in the explorer above to compare.

Explore enforcement modes

04 / YOUR AGENTS

Your agents get to work.

A shell command. A file read. An MCP tool call. AgentShield sits at the point where intent becomes action, alongside the tools your team already uses.

EXAMPLE REQUEST

Claude Code→Shell hook

git status

The command is evaluated before execution.

Agent integrations

Every action. In view.

Example activity

Policy modeEnforce

EvaluationOn device

Audit trailConnected

Recent decisions Shell commands

Command	Agent	Decision
`git status`	Claude Code	Allowed
`cat ~/.ssh/id_rsa`	Cursor	Blocked
`chmod -R 777 /var/www`	Claude Code	Audited

From a local decision to a shared audit trail.Explore the interactive showcase →

Try it

Try a command.
See the reasoning.

Paste a shell command or choose an example. AgentShield checks it against the community rules and explains its decision. No commands are executed.

Shell commandCommunity rules

Explore the full playground →

The verdict, the rule that fired, and why.

For security teams

Your fleet. One clear record.

Fleet. Every agent, its rules, its last check-in.
Policy. Enable a framework; rules reach every machine.
Audit trail. Hash-chained, exportable.
Posture. Which controls real activity exercises.

Open the showcase Handbook

Rules attesting SOC 2 Type IIExample evidence

Rule	Decision	Fires	Control
sec-block-git-log-credential-mining	BLOCK	59	CC6.1 · CC6.7
ai-mcp-tool-description-poisoning	BLOCK	51	CC6.8 · CC7.2
sec-df-block-cred-to-network	BLOCK	40	CC6.7
ai-llm-output-code-eval-python	AUDIT	52	CC7.2 · CC8.1
ne-block-dns-exfil-backtick	BLOCK	21	CC6.7

The problem

Powerful agents need boundaries.

Agents inherit your permissions. Their actions deserve the same visibility and controls as the rest of your stack.

No exploit required

One poisoned tool description or README is enough to send a private key to a stranger. How →

Check the action itself

EDR sees a trusted shell. The firewall sees HTTPS. AgentShield checks the command before it becomes an action.

A record you can stand behind

"How do you control what your AI agents can do? Show me the record." Most teams can't. What that looks like →

Install to enforced

Install once.
Work with confidence.

Four commands on a laptop, one switch in the dashboard.

Read the setup guide →

Step 1

Install AgentShield

One Homebrew command. Prebuilt binary, macOS and Linux.

$ brew tap AI-AgentLens/tap && brew install --cask agentshield

Step 2

Hook your agent

Claude Code, Cursor, Windsurf, Codex CLI, Gemini CLI, OpenClaw, or the MCP proxy.

$ agentshield setup claude-code

Step 3

Connect your team

Device-code sign-in. Policy and rule packs pull down; decisions stream up.

$ agentshield login

Step 4

Watch for a week, then enforce

Read what would have been blocked. Then flip the organisation to enforce.

$ agentshield scan

Built with clear boundaries

Clear about what we do.

Checks, without execution.

The agent asks "may I?". We answer.

Evaluation stays local.

No model calls or telemetry in the local evaluator. Connect to Agent Lens when you want a shared audit trail.

Open about its limits.

We publish our own bypasses. Start here →

Pricing

Start local. Scale with your team.

Start with visibility. Add enforcement and longer retention as your team grows.

Free

5 members, 5 agents
OWASP LLM Top 10
Audit-only mode
7-day retention

Start free

Team

$79USD per user / month

$59/user/month billed annually
5-user minimum · model usage paid separately
Up to 50 members, 100 agents
Observe, Assist, Limited Autopilot
OWASP, SOC 2, ISO 27001
Enforcement, custom rules
90-day retention, reports

Request Team access

Enterprise

Custom

Contracted team and fleet capacity
Adds GDPR and HIPAA
Custom data labels
1-year retention, priority support

Talk to us

Get started

Your next agent.
With a clear boundary.

Install on your own machine first. Then invite the team. [email protected]