Initialize an empty pipeline.

Initialize the exporter with formatting options. Args: pretty_print: Whether to format output for readability.

Auto-trigger LLM generation after expert submits.

Run after a task is approved: refresh VALID_TASK_IDS, domains.json, skill.md.

Read and return the account_setup.json contents.

Load and return the action_items.json contents.

Load the actions log.

Add a transformation step to the pipeline. Args: func: A callable that takes data and returns transformed data. name:

Admin generates invite codes (no expert account needed).

Aggregate records by a grouping field. Groups records by the specified field and applies an aggregation function to the aggregation field wit

Load the aggregated results.

Collect all step contents across all jobs.

Raw text for broad pattern matching.

Load and return the policy analysis JSON.

Load and return the analytics.json contents.

Return the announcements directory path.

Read and return the generated answer.txt contents.

Import api_client.py from the workspace.

Create a lookup dict by application ID.

Import the app package from the workspace.

Calculate the discounted price. Args: percent: Discount percentage (0-100). Returns: The price after discoun

Approve a generated task and write files to the tasks/ directory. Optionally accepts edited file contents to override the generated ones.