The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to analyze potentially untrusted data from various external sources—including support tickets, community threads, sales notes, and web research on platforms like GitHub and Hugging Face—to generate its segment maps. This creates a surface for indirect prompt injection where instructions hidden within that data could attempt to influence the agent.
Ingestion points: Web browsing for market sizing data, local files (devrel-context.md), and raw text inputs provided during the strategic interview phase (e.g., support ticket contents).
Boundary markers: The instructions do not define specific delimiters or directives to ignore instructions that might be embedded in the analyzed data.
Capability inventory: The skill primarily generates text-based strategic documents and lacks high-risk capabilities such as arbitrary code execution, file system modifications (beyond context storage), or network exfiltration of sensitive files.
Sanitization: No explicit sanitization or filtering logic is prescribed for the ingested external content.
[EXTERNAL_DOWNLOADS]: The skill directs the agent to perform research on established and well-known platforms, including Stack Overflow, GitHub, RedMonk, Hugging Face, and Kaggle. These references are used for legitimate market sizing and data aggregation purposes essential to the skill's primary function.

developer-segmentation