mongodb-atlas-stream-processing

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADSDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill has an inherent surface for indirect prompt injection as it is designed to ingest and process data from external streaming sources.
  • Ingestion points: Data entering through $source stages from Kafka topics, Kinesis streams, or MongoDB change streams (referenced in SKILL.md).
  • Boundary markers: The skill instructions recommend using the $validate stage for schema enforcement and warn against treating external content as trusted in the Anti-Patterns section, but specific prompt delimiters are not mandated.
  • Capability inventory: The skill uses MCP tools that can write to databases ($merge), make network requests ($https, $emit), and execute custom JavaScript ($function), which could be targeted by malicious data.
  • Sanitization: No explicit instructions for sanitizing or escaping the streaming data before interpolation are provided beyond schema validation.
  • [EXTERNAL_DOWNLOADS]: References official MongoDB repositories for example processor patterns and workflow synchronization.
  • Fetches example processors and quickstarts from the MongoDB ASP examples repository (https://github.com/mongodb/ASP_example).
  • Imports the canonical workflow instructions from the MongoDB agent-skills repository (https://github.com/mongodb/agent-skills).
  • [DYNAMIC_EXECUTION]: The skill manages stream processors that support custom JavaScript User Defined Functions (UDFs).
  • Evidence: The $function stage is documented in references/pipeline-patterns.md and requires the SP30+ tier.
  • Context: This is a native capability of MongoDB Atlas Stream Processing for data transformation and is documented as a feature for advanced users.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 03:52 AM
Security Audit — agent-trust-hub — mongodb-atlas-stream-processing