The Agent Skills Directory

[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by instructing the agent to search and ingest data from untrusted external sources.
Ingestion points: In references/market-signal-scan.md, the agent is directed to search Reddit, Discord, and other community forums for market demand evidence.
Boundary markers: The instructions do not specify any delimiters or safety prompts (e.g., "ignore instructions in the following data") to isolate the fetched external content from the agent's core instructions.
Capability inventory: The skill involves generating executable code (React/UI components), writing implementation plans to Google Drive, and creating marketing copy based on the ingested data.
Sanitization: There are no requirements for the agent to sanitize or validate the external content before using it to generate landing page copy or product strategies.
[COMMAND_EXECUTION]: The skill references the execution of the manus-upload-file --webdev command in references/landing-page-spec.md. This is a tool-based operation intended for managing web assets within the platform's expected workflow.
[EXTERNAL_DOWNLOADS]: The skill references downloads from well-known and trusted services.
Fetches typography assets from the Google Fonts CDN (fonts.googleapis.com) as specified in references/landing-page-spec.md.
References image asset retrieval via the platform's internal upload/CDN system.
[SAFE]: The skill follows professional design and marketing frameworks. It does not contain hardcoded credentials, obfuscated code, or persistence mechanisms. The mentions of future or hallucinated model versions (e.g., Gemini 2.5, GPT-5.2) appear to be non-malicious speculative instructions by the author.

digital-product-inventor