The Agent Skills Directory

[SAFE]: No malicious patterns or security vulnerabilities were detected. The skill promotes industry-standard security practices for data integration.
[PROMPT_INJECTION]: The skill provides architectural patterns for ingesting data from external SaaS platforms and event streams, which constitutes a surface for indirect prompt injection. This is mitigated through documented best practices.
Ingestion points: External data entering via webhooks and API responses as described in references/enterprise-connectors.md.
Boundary markers: Recommends using signature verification (e.g., for Stripe webhooks) and idempotency keys to validate source authenticity.
Capability inventory: Data processing is handled through standard libraries like requests and dlt for warehouse ingestion.
Sanitization: Emphasizes the use of schema contracts and Pydantic validation to enforce data integrity and prevent schema confusion.
[CREDENTIALS_UNSAFE]: The skill explicitly advises using environment variables and secrets managers for API keys and tokens, avoiding hardcoded credentials in code snippets.
[EXTERNAL_DOWNLOADS]: Installation and reference links point to the author's official GitHub repository, which is a trusted vendor resource for this skill.
[REMOTE_CODE_EXECUTION]: No patterns for arbitrary command execution or remote script execution from untrusted sources were found. Package requirements are limited to standard and safe data engineering libraries.

data-integration