The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses the twenty CLI (via yarn twenty) to perform operational tasks such as managing remotes, syncing changes, building, and deploying. These actions are standard for the platform and within the intended scope of the skill.
[CREDENTIALS_UNSAFE]: The instructions correctly advise against hardcoding API keys in source files and demonstrate the use of environment variables ($TWENTY_API_KEY) and repository secret stores for CI/CD workflows.
[PROMPT_INJECTION]: The skill reads project files such as package.json and src/application-config.ts and monitors CLI output. While this creates a surface for indirect prompt injection from application data, it is a necessary requirement for the skill's primary purpose of troubleshooting and managing existing codebases.
[DATA_EXFILTRATION]: The skill interacts with local configuration in ~/.twenty/config.json through its own CLI to manage authenticated remotes. No patterns suggestive of unauthorized data exfiltration to external domains were detected.

manage-app