The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses a set of specialized MCP tools (atlas-streams-discover, atlas-streams-build, atlas-streams-manage, atlas-streams-teardown) to perform administrative actions on MongoDB Atlas cloud resources. These operations are within the stated scope of the skill and include mandatory user confirmation steps for destructive actions.
[EXTERNAL_DOWNLOADS]: The instructions direct the agent to fetch validated pipeline patterns and quickstarts from the official MongoDB Atlas Stream Processing examples repository on GitHub (https://github.com/mongodb/ASP_example). This is a trusted source used for configuration guidance.
[PROMPT_INJECTION]: No malicious instruction overrides or bypass attempts were detected. The skill contains 'CRITICAL' and 'IMPORTANT' markers, but these are used legitimately to enforce safety protocols, such as mandatory connection validation before processor creation.
[CREDENTIALS_UNSAFE]: The skill documentation includes examples of connection configurations (e.g., Kafka, HTTPS, S3) that use standard placeholders for sensitive fields like apiClientSecret, password, and Authorization headers. It explicitly instructs the agent to use MCP elicitation to collect these from the user rather than hardcoding them.
[DATA_EXFILTRATION]: While the skill manages data movement (sinks and sources), all operations are confined to the user's Atlas project and configured connections (Kafka, S3, etc.). There are no patterns suggesting unauthorized exfiltration of data to external attacker-controlled domains.

atlas-stream-processing