The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill recommends installing official packages from the Dagster ecosystem, such as dagster, dagster-webserver, and various integrations including dagster-dbt, dagster-snowflake, and dagster-aws. These are established, official packages for the platform.
[COMMAND_EXECUTION]: The documentation provides examples for using Dagster Pipes to execute computation in external environments including subprocesses, Kubernetes pods, and Docker containers. These represent standard architectural features of the platform for running workloads in isolated environments.
[INDIRECT_PROMPT_INJECTION]: The skill documents ingestion points for external data from URLs and APIs which are subsequently processed in pipelines. This includes examples where this data is used in LLM completions for ticket classification.
Ingestion points: 00-overview.md (CSV URL), 02-resources-io-managers.md (GitHub API), 12-ai-ml-pipelines.md (training data).
Boundary markers: Absent in instructional code snippets.
Capability inventory: 08-dagster-pipes.md (subprocess execution), 03-ops-jobs-graphs.md (op execution), 02-resources-io-managers.md (file system writes).
Sanitization: Absent in simplified documentation examples.
[DYNAMIC_EXECUTION]: The deployment and Pipes references describe launching user code via gRPC servers and external process clients. These patterns are essential to the framework's operation and are documented using secure configuration methods like dg.EnvVar for sensitive data.

dagster