skills/spotdraft/oncall-investigator-skills/contract-computation-null-propagation/Gen Agent Trust Hub
contract-computation-null-propagation
Pass
Audited by Gen Agent Trust Hub on Apr 9, 2026
Risk Level: SAFEDATA_EXFILTRATIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [DATA_EXFILTRATION]: The skill fetches contract metadata and questionnaire state from the vendor's internal BigQuery datasets (
spotdraft-prod,spotdraft-qa) and administrative endpoints (e.g.,api.in.spotdraft.com/admin/). This access is localized to the vendor's infrastructure for support purposes. - [COMMAND_EXECUTION]: Provides SQL queries for execution via BigQuery MCP tools to inspect contract data stored in
public_contracts_v3_contractdatatables. - [PROMPT_INJECTION]: Contains a vulnerability surface for indirect prompt injection (Category 8). The agent is instructed to ingest and analyze untrusted data (ContractData and Liquid template definitions) from external databases.
- Ingestion points: BigQuery SQL results from
spotdraft-prod.prod_india_db.public_contracts_v3_contractdataand API responses fromapi.{cluster}.spotdraft.com. - Boundary markers: None explicitly defined in the provided instructions to separate untrusted data from processing logic.
- Capability inventory: Execution of SQL queries via MCP tools and network access to administrative URLs.
- Sanitization: No specific sanitization or validation of the retrieved contract data is mentioned before analysis.
Audit Metadata