What is AgentSecrets?
The Zero-Knowledge Difference
How AgentSecrets Works
Installation
Quick Start
Migrating from .env Files
Migrating from Vault / AWS
Migrating from dotenv-vault
Production Checklist
Credential Exposure
What Zero-Knowledge Means
The Proxy Model
The Three-Layer Model
Environments
Agent Identity
Storage Modes
The No get() Principle
Secret-Level Policies
Cloud Overview & Architecture
The Dual-Engine Model
Cloud Resolver Data Plane
Workload & Agent Tokens
Egress Allowlists & Audit Streams
Cloud REST API Reference
Account (init / login)
Server & Self-Hosting (server)
Docs
Shell Autocompletion
Keychain Auth
Secrets
Environments
Credential Proxy
env Injection
Workspaces & Teams
Projects
Agent Identity
Audit & Governance
Integrations Overview
Claude Desktop
Cursor
OpenClaw
HTTP Proxy (Any)
LangChain (Soon)
CrewAI (Soon)
CI/CD Pipeline
SDK Overview
Python SDK
Python API Reference
Python SDK Manual Testing
JavaScript SDK (Soon)
Ecosystem Overview
Zero-Knowledge MCP Server
Server Overview
5-Layer Architecture
Self-Hosting Guide
Authentication & Keys
Workspaces & Teams
Projects & Scope
Environments
Secrets & Sync Protocol
Agent Identity Resolution
Telemetry & Metrics Engine
Audit Log Sync
API Endpoint Reference
Security Overview
Anti-Impersonation & Process Verification
Encryption Model
Zero-Knowledge Sync
Proxy Security Layers
Threat Model
OWASP Top 10 Mitigation
Security FAQ
Third-Party Audit
Reporting Vulnerabilities
Guides Overview
Building on the SDK
Stripe Integration
OpenAI Integration
Multi-Agent Setup
Onboarding Team
CI/CD Pipeline
Publishing ZK MCP
Rotating Credentials
Auditing Team Activity
Dev to Production
Kubernetes Deployment
Monorepo Setup
Production Proxy Hardening
vs .env Files
vs HashiCorp Vault
vs AWS Secrets Manager
vs dotenv-vault
vs Infisical
When Not to Use
Proxy Not Starting
Proxy Not Resolving
Domain Blocked
Sync Conflicts
MCP Not Connecting
Session Token Errors
Proxy Session Authorization
Keychain Storage & Backends
SSRF & Destination Rules
Installation Issues
Error Codes Reference
Frequently Asked Questions
v3.1.x
v3.0.0
v2.1.0
v2.0.0
v1.4.0
v1.3.x
v1.2.0
v1.1.x
v1.0.x
AgentSecrets CloudCloud Overview & Architecture

AgentSecrets Cloud Overview & Architecture

AgentSecrets Cloud is the zero-knowledge credential data plane for distributed workloads, CI/CD pipelines, and autonomous AI agents in production.

While the local AgentSecrets CLI secures credentials on developer workstations using native OS Keychains (macOS Keychain, Linux Secret Service, Windows Credential Manager), AgentSecrets Cloud extends these zero-knowledge guarantees to cloud environments: resolving credentials in volatile, non-swappable RAM on the wire, enforcing workspace-wide egress domain allowlists, and streaming tamper-evident audit logs across your fleet.


The Security Problem in Cloud Deployments

When deploying AI agents or microservices to cloud environments (Docker, AWS ECS, Render, Vercel, Kubernetes), the operating system's native keychain is unavailable.

Traditionally, teams are forced to fall back on passing plaintext API keys as environment variables. In modern AI architectures, this creates severe vulnerability surfaces:

  • Prompt Injection & Context Exfiltration: An autonomous agent processing untrusted user input can be manipulated into outputting its environment variables or secret strings into chat responses or LLM prompt context.
  • Supply Chain & Dependency Compromise: Third-party packages (via NPM, PyPI, or Go modules) running in the same process space can inspect process.env or read local disk files to exfiltrate keys.
  • Observability & Error Leakage: Unhandled exceptions, stack traces, and debug logs frequently reflect raw authorization headers into centralized logging systems (Datadog, CloudWatch, Sentry).

AgentSecrets Cloud eliminates these failure modes by keeping plaintext credentials completely outside the application runtime.

Loading diagram...

Core Architectural Guarantees

1Ephemeral In-Memory Execution

Decrypted workspace encryption keys and credential values never touch disk, swap space, or persistent storage. Decrypted key material exists solely in volatile memory buffers locked with mlock(2) system calls via memguard. Memory buffers are immediately overwritten with cryptographic zeros upon process termination.

2Client-Encrypted Delegation (CEDK)

AgentSecrets Cloud utilizes X25519 asymmetric cryptography and NaCl SealedBoxes. Your workspace master Data Encryption Key (DEK) is encrypted client-side on your workstation before upload. The central control plane stores opaque ciphertext; only the isolated Cloud Resolver unseals the DEK in temporary RAM during authorized execution.

3Egress Domain Allowlist Firewall

Before any credential placeholder is resolved, the Cloud Resolver evaluates the outbound target URL against your workspace allowlist using sub-microsecond radix tree matching. If an agent attempts to route a credential to an unauthorized host, the request is terminated immediately with 403 Forbidden (domain_blocked), and no secret is ever decrypted or sent.

4Active Response Redaction

Upstream API responses are scanned in real time across multiple encoding variants (Plaintext, Base64, Hex, URL-encoded, and JSON-escaped). If an upstream provider echoes a credential in an error message or response payload, the scanner replaces the value with [REDACTED] before returning data to the calling agent.


Execution Modes in the Cloud

AgentSecrets Cloud supports three execution models depending on your infrastructure:

1Container Startup Ingestion ()

Runs your application or container entrypoint directly:

agentsecrets env -- node server.js

The CLI authenticates using a scoped Workload Token (AGENTSECRETS_TOKEN), retrieves project secrets over TLS, and injects them directly into child process memory. No .env files are written to disk.

2Python SDK & Transparent Proxy

Integrates directly into your application codebase:

from agentsecrets import AgentSecrets, credential secrets = AgentSecrets() response = secrets.call( "https://api.stripe.com/v1/charges", method="POST", headers={"Authorization": f"Bearer {credential.STRIPE_KEY}"}, body={"amount": 2000, "currency": "usd"} )

The SDK passes the placeholder reference (credential.STRIPE_KEY). The Cloud Resolver injects the real key on the wire over TLS and returns the clean API response.

3Direct REST Resolution ()

Enables serverless functions (AWS Lambda, Cloudflare Workers) and custom tool-calling agents to resolve credentials via standard HTTP requests without running a local proxy daemon.


Local Engine vs. AgentSecrets Cloud

CapabilityLocal Engine (Workstation)AgentSecrets Cloud (Production)
Primary EnvironmentDeveloper Laptops (macOS, Linux, Windows)Containers, VMs, Serverless, AI Agents
Secret StorageLocal OS Keychain (Apple Keychain, Secret Service)Ephemeral Volatile RAM via Client-Encrypted DEK
Connection to AgentSecrets servers?Not required (Resolves locally from OS Keychain)Required (Connects to Cloud Resolver)
AuthenticationLocal login sessionScoped Workload Tokens (agt_prod_...)
Audit LoggingLocal SQLite fileCentralized real-time log stream on Web Dashboard
Domain FirewallLocal allowlistCentral workspace domain allowlist

Next Steps

Was this helpful?
Thanks for your feedback!
Your feedback helps us improve the platform.