The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes unvalidated string interpolation (f-strings) to construct SQL queries within its examples and scripts, which is an insecure practice vulnerable to SQL injection if variables are sourced from untrusted input.
Evidence: engine.execute(f"SELECT status FROM etl_runs WHERE run_id = '{run_id}' AND run_date = '{run_date}'") in EXAMPLES.md (lines 35-40).
Evidence: query = f"SELECT * FROM {table_name}" used in dynamic task creation in REFERENCE.md (line 80).
[DATA_EXFILTRATION]: The provided ETL framework and reference examples include logic for moving data from internal sources to external sinks, such as cloud storage buckets or remote databases.
Evidence: The _load_s3 function in scripts/run_etl_pipeline.py uses boto3 to upload data to S3 buckets.
Evidence: Snowflake data loading operations using SnowflakeHook and SnowflakeOperator are detailed in REFERENCE.md.
[PROMPT_INJECTION]: An indirect prompt injection surface is present because the skill is designed to ingest and process data from external sources while having the capability to execute database commands and perform network operations.
Ingestion points: The skill ingests untrusted data via requests.get (APIs) and pd.read_sql (External Databases) in scripts/run_etl_pipeline.py and REFERENCE.md.
Boundary markers: Absent. No delimiters or instructions are provided to the agent to distinguish between data and embedded instructions during processing.
Capability inventory: The skill has the ability to execute SQL queries via SQLAlchemy and Snowflake providers, write files to S3 via Boto3, and perform arbitrary network requests via the requests library.
Sanitization: Absent. SQL commands are constructed using raw string interpolation without the use of parameterized queries or escaping mechanisms.

data-engineer