12% of MCP Configuration Files Contain Hardcoded Secrets Threatening Enterprise Security
News related to:Hush Security · 2 min read
Hush Security, a company specializing in securing non-human identities, has released a comprehensive analysis revealing a significant security risk in the Model Context Protocol (MCP) configuration files used by enterprises. The research, titled "The State of MCP Configuration," examined approximately 82,000 public MCP configuration files, uncovering a concerning trend: 12% of the credential slots in these files contain hardcoded secrets, and 55% of these secrets lack recognizable patterns that secret-scanning tools can detect.
According to Hush Security's findings, the MCP configuration files, which are designed to be committed to source control for team collaboration, are accumulating a new class of machine identity risks. These files tell AI agents which tools to connect to and how to authenticate, but the design choice to commit these files to source control comes with a significant security downside. The research highlights that the secrets within these files are often broad-scope and non-expiring, posing a substantial threat to organizations.
Among the leaked credentials, 53% have a defined scope that affects entire organizations, accounts, workspaces, or databases. Moreover, 80% of these credentials never expire by default, making them a persistent threat. Hush Security's analysis further revealed that 24% of the hardcoded secrets in the dataset are both broad-scope and non-expiring, creating a particularly dangerous situation for enterprises.
Micha Rave, CEO and Co-Founder of Hush Security, emphasized the urgency of addressing this issue. "When it is, the highest-risk credentials in them match no known pattern, and the identities behind them have no owner and no expiry. That's a whole population of access tokens sitting out in public Git with no one watching."
Hush Security recommends four immediate steps for security and engineering teams managing MCP configuration files:
1. Never commit an inline secret; use variable expansion wherever possible. 2. Move off static, long-lived credentials entirely in favor of short-lived, identity-based access. 3. Where short-lived credentials are not yet possible, rotate any secret that was ever committed. Deleting the line doesn't remove it from history. 4. Keep credentials out of the agent and the MCP. Broker them from a secure location so there's nothing to steal at the edge. 5. Give every agent identity an explicit owner and an expiry.
The full report, including methodology, is available at https://www.hush.security/state-of-mcp/.
This research underscores the growing importance of securing machine identities in the age of agentic AI. Enterprises must take proactive measures to mitigate the risks associated with hardcoded secrets in MCP configuration files to protect their sensitive data and maintain compliance.