The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands using the Symfony console (bin/console), Composer, and DDEV (ddev exec) for project management, debugging, and tooling. These are standard practices for local PHP development.
[DATA_EXFILTRATION]: The instructions reference commands that expose sensitive application data to the terminal for debugging purposes. Specifically, references/debugging.md mentions bin/console secrets:list --reveal to display encrypted environment variables and bin/console dbal:run-sql for executing arbitrary database queries.
[EXTERNAL_DOWNLOADS]: The skill facilitates the download of third-party code through standard ecosystem tools. It uses composer require for PHP packages and ddev add-on get for DDEV service extensions (e.g., Redis, RabbitMQ) as documented in SKILL.md and references/ddev.md.
[INDIRECT_PROMPT_INJECTION]: The skill exposes a surface for indirect prompt injection by having the agent ingest untrusted data from the user's project files.
Ingestion points: The agent reads and analyzes composer.json, .ddev/config.yaml, and project directories (src/, config/) to determine the Symfony version and project configuration.
Boundary markers: The instructions do not define explicit delimiters or 'ignore embedded instructions' warnings for content read from the project filesystem.
Capability inventory: The agent possesses capabilities to execute shell commands (ddev, bin/console) and write code files using scaffolding tools (make:controller, make:entity).
Sanitization: No sanitization or validation of the content found within project files is described.

symfony