OneCLI: Open-Source Credential Vault for AI Agents
What OneCLI Solves
AI agents are frequently given raw API keys for accessing external services, creating security risks. OneCLI addresses this by acting as a credential vault that sits between agents and the services they call. Instead of baking API keys into every agent, you store credentials once in OneCLI's encrypted vault and give agents placeholder keys (like FAKE_KEY).
How It Works
When an agent makes an HTTP call through the OneCLI proxy, the gateway matches the request by host and path patterns, verifies the agent should have access, swaps the placeholder for the real credential, and forwards the request. The agent never touches the actual secret—it just uses CLI or MCP tools as normal.
Technical Architecture
- Rust Gateway: Fast HTTP gateway that intercepts outbound requests and injects credentials. Agents authenticate with access tokens via Proxy-Authorization headers.
- Web Dashboard: Next.js app for managing agents, secrets, and permissions (port 10254).
- Secret Store: AES-256-GCM encrypted credential storage. Secrets are decrypted only at request time.
- Embedded Database: Runs with embedded PGlite (PostgreSQL-compatible) or can use external PostgreSQL.
Quick Start
Run locally with Docker:
docker run --pull always -p 10254:10254 -p 10255:10255 -v onecli-data:/app/data ghcr.io/onecli/onecliThen open http://localhost:10254, create an agent, add your secrets, and point your agent's HTTP gateway to localhost:10255.
Key Features
- Transparent credential injection: agents make normal HTTP calls
- Encrypted secret storage with AES-256-GCM encryption at rest
- Host and path pattern matching for routing secrets to specific API endpoints
- Multi-agent support with scoped permissions per agent
- No external dependencies in single-container mode
- Two auth modes: single-user (no login) for local use, or Google OAuth for teams
- Apache-2.0 licensed
Compatibility
Works with any agent framework (OpenClaw, NanoClaw, IronClaw, or anything that can set an HTTPS_PROXY). The project is structured with the Rust proxy on port 10255 and Next.js dashboard on port 10254.
📖 Read the full source: HN AI Agents
👀 See Also
Using Claude to audit OpenClaw setup reveals security issues
A developer used Claude to review their OpenClaw installation and discovered the bot was writing API keys in clear text in memory and JSON files, along with other security concerns.
AI System Discovers 12 OpenSSL Zero-Days, Curl Cancels Bug Bounty Due to AI Spam
AISLE's AI system discovered all 12 zero-day vulnerabilities in OpenSSL's recent security release, marking the first large-scale demonstration of AI-based cybersecurity. Meanwhile, curl cancelled its bug bounty program due to AI-generated spam submissions.
Claude Android App Reportedly Reads Clipboard Without Explicit User Action
A user reports that the Claude Android app analyzed code from their clipboard without them pasting it, with Claude identifying the file as pasted_text_b4a56202-3d12-43c8-aa31-a39367a9a354.txt. The behavior couldn't be reproduced in subsequent tests.
Zero-Trust OpenClaw Architecture Adds Pre-Execution Authorization and Post-Execution Verification
An open-source architecture for OpenClaw adds two security checkpoints: a Rust sidecar that intercepts tool calls before execution with sub-millisecond authorization overhead, and deterministic post-execution verification using assertions instead of LLM judgment. The system includes tracing with DOM snapshots and screenshots, plus a DOM compression skill that reduces token usage by 90-99%.