Infisical/agent-vault

Health v2 · maintenance only

Flags: young no_license

Adoption not part of the score

2141 stars · 139 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

Agent Vault is an open-source HTTP credential proxy and vault by Infisical that sits between AI agents and the APIs they call. It stores credentials server-side and injects them into outbound HTTP requests via an MITM proxy, so agents like Claude Code, Cursor, Codex, OpenClaw, and custom harnesses never hold real secrets.

Use cases

• prevent AI agents from leaking API keys via prompt injection
• let Claude Code call Anthropic and GitHub without holding credentials
• run coding agents in sandboxes with brokered API access
• route custom agent harness HTTP traffic through a credential-injecting proxy
• back agent credentials with an external secrets store like Infisical
• secure ephemeral CI or VPS agent environments against credential exfiltration

When to choose

• you run AI agents that make outbound API calls and worry about prompt-injection-driven secret theft
• you want agents to use dummy credentials while a proxy swaps in real ones
• you need a self-hosted, single-binary vault plus MITM proxy for agent workloads

When to avoid

• you only need traditional secrets management for human-operated services without agents
• your traffic cannot be routed through an HTTPS proxy or MITM architecture
• you need a fully managed SaaS solution rather than self-hosting

Facets

service · maturity active

secrets-management proxy security http-server cli agent-framework security self-hosted developer-tools large-language-models self-hosted cli go credential-broker mitm-proxy credential-exfiltration ai-agent-security https-proxy infisical ai-agents linux macos docker

Member repositories

For agents

markdown · JSON · MCP: product_card(name="Infisical/agent-vault")