The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from external sources such as APIs, databases, and files (e.g., CSV, JSON). This creates a potential surface for indirect prompt injection if external data contains malicious instructions.
Ingestion points: The extractors section in SKILL.md defines data entry from REST APIs, GraphQL, PostgreSQL, MySQL, MongoDB, and file sources.
Boundary markers: While explicit delimiters for untrusted content are not defined in the templates, the skill includes a data_quality section for schema validation.
Capability inventory: The skill uses database_query, api_fetch, spreadsheet_write, and file_transform tools.
Sanitization: The skill provides templates for cleaning (e.g., remove_nulls), validate against schema, and schema_validation to sanitize and verify external data.
[DYNAMIC_EXECUTION]: The skill includes examples of JavaScript code blocks used for data transformation and aggregation. While this involves dynamic execution, it is the primary intended functionality of an ETL workflow and uses standard logic for type conversion and field mapping.
[EXTERNAL_DOWNLOADS]: The skill configuration references integrations with well-known and trusted services such as Shopify, Stripe, Google BigQuery, Google Sheets, Salesforce, and HubSpot. These references are standard for data integration workflows and do not involve downloading or executing untrusted remote code.

data-pipeline