The Agent Skills Directory

[PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill is designed to ingest untrusted user content to generate outlines and image prompts, which creates a risk of embedded instructions influencing the agent's behavior.
Ingestion points: User-provided markdown content saved to source.md (Step 1.2).
Boundary markers: Absent. Content is directly interpolated into templates defined in references/base-prompt.md and references/outline-template.md without clear delimiters.
Capability inventory: The skill can execute shell commands (python3, npx), write to the local filesystem, and perform network requests.
Sanitization: Absent. The skill does not perform validation or escaping of the user-supplied content.
[EXTERNAL_DOWNLOADS]: The skill uses npx -y bun in Step 8 to run helper scripts. This involves downloading the bun runtime and Node.js packages (pdf-lib, pptxgenjs) from public registries at runtime.
[COMMAND_EXECUTION]: The skill relies on shell command execution for its core workflow.
Uses test -f and test -d to check for configuration files and existing directories (Step 1.1, Step 1.3).
Executes python3 scripts/generate_slide.py to interface with an external image generation service (Step 7).
Executes npx -y bun to run TypeScript scripts for merging files into PDF and PPTX formats (Step 8).
[DATA_EXFILTRATION]: The scripts/generate_slide.py script performs network requests to api.labnana.com using urllib.request. This domain is used to send slide content for image generation. While functional, it is a non-whitelisted external service.

space-slide-deck