Governed AI execution
·
SOC2-ready
·
Open-source core

Execution gatekeeper for AI agents

Policy-driven governance, risk scoring, and human-in-the-loop approval for AI agent tool calls. Built for production with modern web UI, comprehensive CLI, and SDK integrations.

Action Log

shell:run
allow-low-risk
low
http:post
require-approval
high
file:write
block-writes
high
api:call
allow-trusted
medium

High-Risk Action

Requires approval before execution

Tool:http:post
Risk:High
Policy:require-approval
Why Faramesh

Guardrails, auditability, and enterprise integrations

Unified controls across agents, clouds, and internal systems.

Policy-as-Code Guardrails

Centrally manage allow, deny, and require_approval rules to control agent actions in real time.

Audit-Ready Trails

Every action is logged, timestamped, and signed for traceability across teams and environments.

Enterprise Integrations

Connect to your stack—SSO, SIEM, cloud providers, SDKs—without changing developer workflows.

Use cases

Designed for high-trust workflows

Apply the same guardrails to agents, automation, and LLM tools.

Governed AI Agents

Enforce approvals and risk thresholds before agents run sensitive operations.

Secure Automation

Wrap RPA and LLM tools with policy checks, audit trails, and live approvals.

Regulated Workloads

Meet SOC 2 and ISO requirements with immutable logs and least-privilege policies.

Industries

Built for regulated teams

Zero-trust guardrails for finance, healthcare, and public sector workloads.

Financial Services

SOX-aligned approvals and full auditability for trading, treasury, and support bots.

Healthcare

PHI-safe guardrails and zero-trust enforcement for clinical and back-office agents.

Public Sector

Policy-backed automation with strict access controls for sensitive workloads.

Testimonials

Trusted by teams shipping AI in production

Auditability, approvals, and enforcement without slowing down delivery.

"Faramesh gave us the approvals and auditability we needed to ship AI assistants into regulated environments."

Director of Security
Global Financial Platform

"We wrapped every agent action with policy checks in a week. Now every exec gets a clean audit trail."

VP Engineering
Enterprise SaaS
See Faramesh in action

Real-time governance and visibility

Policy-as-code editor, risk dashboard, and comprehensive audit trails.

policy.yaml
rules:
- name:
require_approval
risk_level:
high
effect:
require_approval
Live Preview
Policy Active
• High-risk actions → Approval
• Medium-risk → Auto-allow
• Low-risk → Auto-allow
Approved
4.3M
Pending
142
Denied
23
Risk Trends
Low: 89%
Medium: 8%
High: 3%

Audit Log

Last 24h
shell:runby
agent-1
2m ago
http:postby
admin
5m ago
file:writeby
agent-2
8m ago
api:callby
agent-1
12m ago
Code Examples

Drop-in policy-as-code for your stack

Use SDKs or HTTP to wrap every agent action with approvals and auditability.

Wrap any agent action with governance using the Python SDK

from faramesh import configure, submit_action

# Configure client
configure(base_url="http://localhost:8000")

# Submit action with governance
response = submit_action(
    agent_id="my-agent",
    tool="shell",
    operation="run",
    params={"cmd": "echo 'Hello Faramesh'"}
)

print(f"Status: {response['status']}")
print(f"Action ID: {response['id']}")

# Check if approval is needed
if response['status'] == 'pending':
    print("Action requires approval")
Integrations

Works with your favorite frameworks

One-line governance for LangChain, CrewAI, AutoGen, MCP, and more.

Ready to get started?

Start governing your AI agents today with policy-as-code guardrails.