The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill fetches and analyzes content from external Instagram accounts. Malicious actors could place instructions in Instagram captions or profile bios to influence the agent's report generation or subsequent tasks.
Ingestion points: Data is retrieved from the Apify API in fetch_instagram.py and saved to raw.json.
Boundary markers: Absent. The workflow does not suggest using delimiters (like XML tags) or specific instructions to treat the scraped content as untrusted data.
Capability inventory: The agent can execute local Python scripts, write files to the filesystem, and interact with the Gemini API.
Sanitization: While analyze_posts.py cleans text for keyword extraction, it does not sanitize the content for potential prompt injection patterns before the agent processes it for reporting.
[External Downloads] (LOW): The skill communicates with api.apify.com to scrape Instagram data. This is expected behavior but involves interaction with a third-party service outside the strictly trusted list.
[Data Exposure & Exfiltration] (SAFE): The skill accesses APIFY_TOKEN and GEMINI_API_KEY through environment variables. These credentials are used only for their intended APIs and are not transmitted elsewhere.

instagram-research