The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill exhibits an indirect prompt injection surface due to its data ingestion patterns.
Ingestion points: Untrusted data enters the agent context via the firecrawl (web scraping/search) and octocode (GitHub search) tools as described in the methodology and tools sections.
Boundary markers: Absent. There are no instructions or delimiters provided to help the agent distinguish between its system instructions and potentially malicious instructions embedded in the retrieved web content.
Capability inventory: The skill utilizes tools for data retrieval (scrape, search, githubSearchCode). No high-risk capabilities such as local file system modification or arbitrary command execution were identified in the provided file.
Sanitization: Absent. No logic is defined to sanitize, filter, or escape the content fetched from external sources before it is synthesized into the final report.
[Data Exposure & Exfiltration] (LOW): The skill triggers network operations to non-whitelisted external domains through the use of the firecrawl and octocode MCP tools. This is considered low risk as it is the primary intended function of a research skill and does not involve access to sensitive local files.

research-and-report