The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill establishes a data ingestion pipeline for processing external visitor traits from CDP providers.
Ingestion points: Untrusted data is retrieved from remote CDP API endpoints (e.g., profiles.segment.com) and identity cookies as described in segment.md and audit.md.
Boundary markers: Absent. The skill relies on programmatic mapping logic rather than explicit LLM boundary markers for the data ingestion flow.
Capability inventory: Mapped traits are used to drive personalization through context.update() and handleUniformRoute(), and are serialized into YAML files for deployment via the uniform CLI.
Sanitization: The mapping.md file defines a traitsToQuirks function that performs type validation, casting raw values to strings and applying normalization (trimming and casing) before data is applied to the visitor context.
[SAFE]: The skill correctly manages sensitive credentials by instructing users to store API keys in server-side environment variables and identifies trusted service endpoints for Segment and Uniform.

uniform-cdp-personalization