T09 · Insecure Skill Coding Practices
Warning
- Location
- SKILL.md:59
- Finding
- Credential Exposure Through Command-Line Password Argument## Vulnerability Details **File Location**: `SKILL.md:59` **Vulnerability Type**: Credential exposure through process arguments and command history **Risk Level**: Medium ### Vulnerable Code ```text agent-browser auth save <name> --url <url> --username <u> --password <p> ``` ### Technical Analysis The documented authentication workflow instructs users or agents to pass a password directly through the `--password` command-line argument. When a real credential replaces the placeholder, its plaintext value may become visible in shell history, process inspection interfaces, command-execution logs, agent transcripts, telemetry, or debugging records. This creates an avoidable secret-handling weakness even if the underlying authentication vault stores the credential securely after receiving it. ### Attack Path 1. A user or agent follows the documented command and substitutes a real website password for `<p>`. 2. The shell or agent execution environment records the complete command, or exposes it temporarily through process metadata. 3. A local user, monitoring component, log reader, or party with access to retained transcripts retrieves the plaintext password. 4. The exposed username and password are used to authenticate to the target website. 5. The attacker obtains whatever account access and data are available to the compromised user. ### Impact Assessment Successful exploitation can disclose credentials for authenticated websites. The resulting privileges are limited to those granted to the compromised account, but may include access to private information, account settings, authenticated actions, or administrative functions when a privileged account is affected. Exploitation generally requires access to local process information, shell history, execution logs, telemetry, or agent transcripts containing the invoked command. The affected scope includes environments where operators follow t ...[truncated 34 chars]
- Remediation
- ## Remediation Suggestions - Replace the command-line password option with an interactive hidden prompt or protected standard-input mechanism. - Support retrieval from an approved secret manager or operating-system credential store without placing the secret in process arguments. - Explicitly warn users and agents never to include passwords, tokens, or other secrets directly in command-line arguments. - Ensure command logging, telemetry, debugging output, and agent transcripts redact credential values. - If compatibility requires `--password`, support references to protected secret sources rather than accepting plaintext values. - Review and remove any existing shell histories, logs, or transcripts that may contain credentials entered using this workflow, and rotate affected credentials.