T09 · Insecure Skill Coding Practices
Warning
- Location
- scripts/generate_auth_token.py:27
- Finding
- Reusable API Credentials Exposed Through Command-Line Arguments and Standard Output<![CDATA[ ## Vulnerability Details **File Location**: `scripts/generate_auth_token.py:27-42` **Additional Locations**: `SKILL.md:43`; `scripts/tomoviee_img2video_client.py:143-148` **Vulnerability Type**: Reversible credential disclosure through process arguments and stdout **Risk Level**: Medium ### Vulnerable Code ```python credentials = f"{app_key}:{app_secret}" access_token = base64.b64encode(credentials.encode()).decode() return access_token if __name__ == "__main__": if len(sys.argv) != 3: print("Usage: python generate_auth_token.py <app_key> <app_secret>") sys.exit(1) app_key = sys.argv[1] app_secret = sys.argv[2] token = generate_access_token(app_key, app_secret) print(f"Access Token: {token}") print(f"\nUse in Authorization header as: Basic {token}") ``` The documented invocation in `SKILL.md:43` is: ```bash python scripts/generate_auth_token.py YOUR_APP_KEY YOUR_APP_SECRET ``` The main client has the same command-line secret exposure pattern at `scripts/tomoviee_img2video_client.py:143-148`: ```python print( "Usage: python scripts/tomoviee_img2video_client.py " "<app_key> <app_secret> <prompt> <image_url> [resolution] [aspect_ratio]" ) sys.exit(1) app_key = sys.argv[1] app_secret = sys.argv[2] ``` ### Technical Analysis Base64 is reversible encoding rather than encryption. The generated token contains the application key and secret in the form `base64(app_key:app_secret)`. Any party able to read the helper's stdout can decode the value and recover both credentials. The helper also requires the secret as a command-line argument. Depending on the execution environment, command-line arguments may be exposed through: - Shell history - Process listings and process inspection interfaces - Job-runner or orchestration metadata - Debug and audit logs - Agent execution transcripts Printing the complete Basic authorization value is unnecessary for the Skill's declared image-to-video oper ...[truncated 1608 chars]
- Remediation
- <![CDATA[ ## Remediation Suggestions 1. **Remove credential output** - Do not print the generated Basic authorization value. - If confirmation is needed, print only a non-sensitive success message. - Never include the token in exceptions, debug logs, or Agent responses. 2. **Stop accepting secrets through command-line arguments** - Read credentials from a protected secret manager or runtime credential provider. - Alternatively, use environment variables supplied through a secure execution mechanism. - For interactive local use, obtain the secret with `getpass.getpass()` so it is not echoed. - Avoid placing secrets directly in shell commands, because environment variables may also leak when assigned inline or logged. 3. **Keep authorization material short-lived** - Construct the authorization header in memory immediately before making the request. - Avoid retaining the encoded credential as a public or long-lived instance attribute where practical. - Do not write either the raw secret or encoded token to disk. 4. **Update the documentation and CLI** - Replace the insecure Quick Start command in `SKILL.md`. - Remove `<app_secret>` from the command-line interface in both scripts. - Document supported secure credential-loading methods and warn that Base64 does not protect secrets. 5. **Apply operational controls** - Rotate any credentials previously used with the helper in logged or Agent-mediated environments. - Restrict access to historical process, CI, orchestration, and Agent logs. - Redact authorization headers and known credential fields at logging boundaries. ]]>