The Agent Skills Directory

[SAFE]: The skill provides documentation and implementation patterns for the official com.apideck:unify Maven package, sourced from standard repositories.
[SAFE]: Best practices for secret management are enforced by instructing developers to use environment variables (e.g., System.getenv("APIDECK_API_KEY")) rather than hardcoding credentials in source code.
[INDIRECT_PROMPT_INJECTION]: The skill enables the agent to fetch and process data from over 200 third-party platforms (CRMs, HRIS, Accounting, etc.). This creates a potential surface for indirect prompt injection if those external systems contain malicious instructions.
Ingestion points: Data is retrieved through various list and get operations defined in SKILL.md (e.g., sdk.crm().contacts().list()).
Boundary markers: The skill does not explicitly instruct the agent to use delimiters or sanitization when processing retrieved data.
Capability inventory: The SDK provides comprehensive read and write access to numerous external APIs, including file storage and HRIS systems.
Sanitization: While the API layer handles standard data validation, the agent's interpretation of the retrieved content depends on its own safety guardrails.

apideck-java