The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several system utilities and local binaries to perform its testing and QA functions.
It uses pkill and open to manage the application lifecycle during testing.
It employs screencapture and sips (Scriptable Image Processing System) to capture and crop images of the application's menu for visual verification.
It executes a bundled bash script scripts/live_provider_matrix.sh which uses node to process JSON data and perform string redaction.
[DATA_EXFILTRATION]: The skill interacts with the application's local configuration file at ~/.codexbar/config.json, which contains sensitive provider keys.
Mitigation: The skill includes explicit jq and node scripts designed to redact apiKey, secretKey, and cookieHeader values before the data is displayed or processed further.
Mitigation: Configuration backups are created with restricted permissions (chmod 600) to ensure they are not world-readable.
[EXTERNAL_DOWNLOADS]: The skill references documentation and API specification URLs for several well-known AI providers (including MiniMax, Deepgram, and Groq).
These references are used solely for looking up official specifications and do not involve downloading or executing external code.
[SAFE]: The skill follows established security best practices for handling sensitive data within a testing environment, such as the use of quoted heredocs in scripts to prevent injection and the integration of secure secret storage tools like 1Password.

qa-test