The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides numerous shell commands (find, grep, git) and embedded Python scripts for the agent to execute. These commands are used to scan the local repository's file structure and content to identify frameworks, locales, and vocabulary. The execution is limited to the local environment and targeted at the project being analyzed.- [PROMPT_INJECTION]: The skill ingests untrusted data from the repository's source code and documentation, which constitutes an indirect prompt injection surface.
Ingestion points: Project files including README.md, package.json, pubspec.yaml, models.py, and *.arb localization files.
Boundary markers: Absent; the skill does not specify the use of delimiters when processing external project data.
Capability inventory: The skill performs file reads and writes results to the local .dev-agent/ directory using shell and Python execution.
Sanitization: Absent; content is extracted and merged for documentation without specific sanitization of instructions that might be embedded in the code or strings.- [DATA_EXFILTRATION]: While the skill reads sensitive configuration files (e.g., .env.example, settings.py) and source code to identify PII logging patterns, it does not include instructions to transmit this data over the network. All outputs are saved locally to the .dev-agent/ folder within the project repository.

setup