The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes user-controlled project files and logs to determine upgrade steps and detect issues. This represents a potential injection surface if those files contain malicious instructions.
Ingestion points: Project metadata in composer.json and deprecation logs in var/log/dev.log as mentioned in SKILL.md and references/deprecations.md.
Boundary markers: No explicit boundary markers or safety instructions are defined to separate user-provided content from agent instructions.
Capability inventory: The agent is instructed to execute shell commands and Docker containers, which could be misused if instructions were successfully injected.
Sanitization: No sanitization of project file content is mentioned before processing.
[COMMAND_EXECUTION]: The skill guides the user and agent to perform system checks and project management using command-line tools.
Evidence: Instructions use commands like php bin/console about, php bin/console debug:container --deprecations, and composer update.
[DYNAMIC_EXECUTION]: The skill utilizes a Docker container to execute Rector, a tool for automated code transformation.
Evidence: SKILL.md provides a command to run rector via the jakzal/phpqa Docker image: docker run --rm -v "$PWD":/project ... jakzal/phpqa:1.125.1-php8.4-alpine rector ....
[EXTERNAL_DOWNLOADS]: The skill facilitates the updating of Symfony Flex recipes from an external source.
Evidence: composer recipes:update fetches updated recipe files from the official Symfony recipes repository on GitHub.

symfony-yoandev-upgrade