Security audit
Security checks for vulnerabilities and agentic risk
The skill is mostly an Atlas Admin API helper, but it asks for broad execution/write authority and includes broad triggers and inconsistent capability claims around mutation and export behavior.
Review this skill carefully before installing. Use it only for MongoDB Atlas Admin API documentation and clearly requested Atlas operations, provide narrowly scoped Atlas API credentials, and require explicit human approval before any POST, PUT, PATCH, or DELETE action. Do not rely on it for generic SQL or unrelated database tasks, and clarify the missing Node scripts and export behavior before use.
The activation text says to use the skill for generic database operations, SQL queries, and data storage management, even though the tool is specifically for MongoDB Atlas Admin API browsing/calling. This overbroad trigger scope increases the chance an agent auto-selects the skill for unrelated tasks and then executes privileged exec-based admin API actions in the wrong context.
The skill is described as a browser/query utility, but the documentation also advertises create/modify/delete execution paths through input_params, config_options, and mutation examples. In an agent setting, this mismatch can cause the tool to be selected for broadly phrased database tasks under a read-oriented label, leading to unintended destructive Atlas API operations.
The skill documents output_format as supporting export/save/convert operations, while later claiming the free edition does not support export of API call results. This inconsistency can mislead an agent or user into invoking data export or persistence behavior that the edition is supposed to forbid, weakening trust boundaries and enabling unintended handling of potentially sensitive API output.
The keyword coverage section enumerates a very broad set of trigger phrases, including generic terms like database operations, SQL, storage management, and create/modify/delete support. In agent ecosystems, such broad matching materially raises the risk of over-invocation, causing this exec-capable Atlas admin skill to be chosen for unrelated or ambiguous prompts and expanding the blast radius of mistakes.
The free-version limitations state that exporting API call results is not supported, but earlier sections say output handling can export/save/convert. Contradictory operational guidance is dangerous in agent workflows because policy or edition restrictions may be bypassed by relying on the more permissive instructions, resulting in unintended data retention or exfiltration of Atlas response data.
No suspicious patterns detected.
