update-cli-config
Warn
Audited by Gen Agent Trust Hub on Sep 17, 2026
Risk Level: MEDIUMPROMPT_INJECTIONDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill provides instructions to modify the
approvalModesetting tounrestricted(described as 'yolo mode'). This allows the agent to bypass user confirmation for all tool calls, potentially removing critical safety guardrails. It also enables modification of permission patterns for tool calls. - [DATA_EXFILTRATION]: The skill identifies and facilitates access to
~/.cursor/cli-config.json, which is explicitly stated to containauthInfo(cached authentication information). This creates a risk of credential exposure if the agent is prompted to read or exfiltrate the file's content. - [INDIRECT_PROMPT_INJECTION]: The skill describes a merging mechanism where project-specific
.cursor/cli.jsonfiles override global settings. An attacker could place a malicious configuration file in a repository to influence the agent's behavior. - Ingestion points:
.cursor/cli.jsonfile in the project directory. - Boundary markers: None.
- Capability inventory: File read and write access to the global CLI configuration and permission modification capabilities.
- Sanitization: No sanitization or validation is mentioned for values read from project-level configuration files before they are merged or potentially written back to the global configuration.
Audit Metadata