The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides explicit instructions for the agent to execute a local Python script (scripts/generate_slide_images.py). This is the intended functional mechanism for the skill's slide rendering capability.
[INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection as it ingests untrusted data from external markdown files (--prompts-dir) and interpolates them into a prompt for the Gemini API.
Ingestion points: The extract_prompt_content function in scripts/generate_slide_images.py reads the entire content of user-provided markdown files.
Boundary markers: The script does not utilize explicit delimiters to separate the ingested file content from the system instructions, although it uses a strong SYSTEM_INSTRUCTION to guide the model's behavior.
Capability inventory: The skill has the capability to perform network operations (Gemini API calls) and write files to the local system (--output-dir).
Sanitization: The script performs basic validation by only processing files that start with digits and skips known metadata files, but it does not sanitize the content of the markdown files themselves.
[SAFE]: The skill follows security best practices for secret management by requiring the GEMINI_API_KEY to be provided via environment variables rather than hardcoding credentials. All external dependencies (google-genai, Pillow) are well-known, standard libraries.

slide-renderer