Researchcode agentssupply chain securityreadme injection
AI Coding Agents Leak Files From READMEs
8.9
New research published March 17, 2026, shows attackers can hide malicious instructions in repository README files to trick AI coding agents during project setup. Tests found semantic injections embedded in installation files can cause agents to execute hidden commands and exfiltrate sensitive local files and secrets. The finding signals a new supply-chain risk for developers and toolmakers who must validate README content and limit automated command execution.
Key Points
- 1Show that README-embedded semantic injections cause AI coding agents to execute hidden install commands and leak files
- 2Demonstrate attackers can exfiltrate sensitive local files and secrets via repository setup instructions
- 3Advise developers and toolmakers to validate README content and restrict agent command execution during setup
Scoring Rationale
High practical impact due to a novel, widely applicable attack vector, limited by single-source reporting and shallow public details.
Sources
Public references used for this report.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
Used by DS/ML engineers at top companies
Active Search Campaigns by BudgetEasyHigh CPC Clicks & Poor Landing PagesMediumCampaign ROAS by Attribution ModelHard
250 free problems · No credit card
See all Ad Tech problems
