The Agent Skills Directory

[COMMAND_EXECUTION]: Provides administrative Python code snippets for use in a Django shell and SQL queries for BigQuery. These tools are used to count, inspect, and soft-delete ContractFullTextMetaData records, as well as to trigger ElasticBackfillReSyncRun workflows.
[EXTERNAL_DOWNLOADS]: References reference materials hosted on Google Docs and administrative endpoints on the vendor's official domain (spotdraft.com). These are used for documentation and verification of system states.
[PROMPT_INJECTION]: Identifies an indirect prompt injection surface regarding the processing of external workspace data. \n
Ingestion points: Variable inputs such as {workspace_id}, {cluster}, and {flag_enabled_at} are ingested through the investigation workflow in SKILL.md.\n
Boundary markers: No explicit delimiters or instructions are used to distinguish these untrusted inputs from the agent's core logic.\n
Capability inventory: The skill possesses capabilities to delete database records and initiate batch processing jobs as defined in SKILL.md.\n
Sanitization: No visible escaping or validation of inputs is performed before they are interpolated into the provided Python and SQL snippets.
[SAFE]: The skill follows operational best practices by including a dry-run mode for data deletion and providing specific filters (e.g., status, file type, and date cutoffs) to prevent over-scoped or accidental data modification.

sdc-historical-pdf-reindex