The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to execute arbitrary shell commands for service orchestration and testing, such as starting a Redis server or running a test suite with 'bun test'.
[INDIRECT_PROMPT_INJECTION]: The skill ingests service definitions and lifecycle commands from external configuration files which could potentially contain malicious instructions.
Ingestion points: Service definitions are read from 'fluentci.toml', 'Procfiles', or JSON files supplied to the 'fluentci services apply' command, as noted in the 'references/services.md' file.
Boundary markers: There are no explicit delimiters or instructions provided to the agent to treat content within these configuration files as data rather than instructions.
Capability inventory: The skill possesses significant capabilities, including executing host commands, managing Docker containers, and running processes via Nix/Devbox environments.
Sanitization: The instructions do not mention any sanitization or validation logic to verify the integrity or safety of the commands defined in the ingested configuration files.
[DYNAMIC_EXECUTION]: The skill dynamically generates and executes service environments and processes based on project-specific configurations provided at runtime via various backends like Docker, Nix, and Host.

fluentci-ci-services