Method and device for generating a recommendation for correcting a configuration file in a computing environment
The invention relates to a method and device for generating a recommendation for correcting a configuration file of an item of infrastructure of a computing environment within which services run. The infrastructure is deployed automatically from said configuration file. The method includes obtaining a runtime report that includes a compliance anomaly detected in the configuration file while said service is being run, and presenting an anomaly correction request as input to a previously trained generative artificial intelligence model. The correction request includes a question formulated in natural language and anomaly context information that includes at least the runtime report. The method also includes obtaining as output a configuration file correction recommendation, validating the recommendation for correcting the configuration file, determining a similarity measurement with respect to the context information, comparing with an acceptability threshold, and when the similarity measurement is validated, submitting the configuration file correction recommendation for user evaluation.
1 . A method for generating a recommendation for correcting a configuration file of an item of infrastructure of a computing environment within which services run, said infrastructure having been deployed automatically from said configuration file, said method comprising:
obtaining and storing in memory a runtime report comprising a compliance anomaly detected in the configuration file during execution of a service in the computing environment;
presenting an anomaly correction request as input to a previously trained generative artificial intelligence model of a Large Language Model type, said anomaly correction request comprising a question formulated in natural language and anomaly context information comprising at least the runtime report, and obtaining as output a configuration file correction recommendation that includes a recommendation for correcting the configuration file;
validating the recommendation for correcting the configuration file, comprising determining a similarity measurement between a corrected version of source code of the configuration file contained in the recommendation and an original version of source code of the configuration file and comparing said similarity measurement with an acceptability threshold;
when the similarity measurement is above the acceptability threshold, submitting the configuration file correction recommendation for user evaluation by a user.
2 . The method according to claim 1 , further comprising
extracting attributes from the runtime report, said attributes comprising at least one anomaly type and a configuration file type,
generating a request to obtain said question from a pre-constructed standard question and said attributes,
presenting said request as input to said previously trained generative artificial intelligence model during a first interaction with the previously trained generative artificial intelligence model, and
obtaining said question as output from the previously trained generative artificial intelligence model.
3 . The method according to claim 2 , further comprising extracting keywords from said attributes of the runtime report and obtaining additional context information by querying at least one data table comprising entries associating context information of the computing environment with at least one keyword of said keywords.
4 . The method according to claim 3 , wherein the additional context information belongs to a group comprising at least:
the configuration file that generated the compliance anomaly,
information on a security policy applicable to the computing environment,
one or more examples of previously obtained correction requests/recommendations.
5 . The method according to claim 2 , further comprising obtaining examples of requests to obtain a correction recommendation by querying a reinforcement learning module, said reinforcement learning module being configured to select at least one example of request from said examples of requests to obtain a correction recommendation from a data table relating to a type of anomaly, said data table comprising examples of correction recommendation requests that have resulted in correction recommendations for an anomaly of said type, validated by the user and associated with a score, as a function of said score.
6 . The method according to claim 1 , further comprising prior reading of a document or governance plan, comprising relative information belonging to a group comprising:
information on a format of the runtime report,
information on at least one attribute of the compliance anomaly to be extracted from the runtime report,
information on the similarity measurement and the acceptability threshold to be used to validate the recommendation for correcting the configuration file.
7 . The method according to claim 1 , wherein said submitting the configuration file correction recommendation comprises an update request, or pull request, to update the configuration file in a source code management repository, in accordance with the configuration file correction recommendation.
8 . The method according to claim 1 , further comprising obtaining an evaluation outcome of the configuration file correction recommendation by said user, and, depending on the evaluation outcome that is obtained, triggering or not triggering an update of the configuration file in accordance with the configuration file correction recommendation.
9 . The method according to claim 8 , further comprising, when said triggering of the update of the configuration file occurs, recording a request to obtain a recommendation and the configuration file correction recommendation as an example in a data table relating to a type of anomaly, in association with a score.
10 . A device that generates a recommendation to correct a configuration file of an item of infrastructure of a computing environment within which services run, said infrastructure having been deployed automatically from said configuration file, said device comprising:
one or more circuits configured to execute a method comprising
obtaining and storing in memory a runtime report comprising a compliance anomaly detected in the configuration file during execution of a service in the computing environment;
presenting an anomaly correction request as input to a previously trained generative artificial intelligence model of a Large Language Model type, said anomaly correction request comprising a question formulated in natural language and anomaly context information comprising at least the runtime report, and obtaining as output a configuration file correction recommendation comprising a recommendation for correcting the configuration file;
validating the recommendation for correcting the configuration file, comprising determining a similarity measurement between a corrected version of source code of the configuration file contained in the recommendation and an original version of source code of the configuration file and comparing said similarity measurement with an acceptability threshold;
when the similarity measurement is above the acceptability threshold, submitting the configuration file correction recommendation for user evaluation by a user.
11 . The device according to claim 10 , wherein said one or more circuits comprise
at least one processor; and at least one memory comprising computer program code, the at least one memory and the computer program code being configured to, together with the at least one processor, cause said device to be run.
12 . The device according to claim 10 , wherein said device further comprises a non-transitory computer program comprising program code instructions which, when said program code instructions executed by computer, cause the one or more circuits to execute the method.
13 . A platform that manages a deployment of a computing environment infrastructure using program code instructions, comprising:
a device that generates a configuration correction recommendation to correct a configuration file of an item of infrastructure of a computing environment within which services run, said infrastructure having been deployed automatically from said configuration file, said device comprising:
one or more circuits configured to execute a method comprising
obtaining and storing in memory a runtime report comprising a compliance anomaly detected in the configuration file during execution of a service in the computing environment;
presenting an anomaly correction request as input to a previously trained generative artificial intelligence model of a Large Language Model type, said anomaly correction request comprising a question formulated in natural language and anomaly context information comprising at least the runtime report, and obtaining as output a configuration file correction recommendation comprising a recommendation for correcting the configuration file;
validating the recommendation for correcting the configuration file, comprising determining a similarity measurement between a corrected version of source code of the configuration file contained in the recommendation and an original version of source code of the configuration file and comparing said similarity measurement with an acceptability threshold;
when the similarity measurement is above the acceptability threshold, submitting the configuration file correction recommendation for user evaluation by a user.
14 . The platform according to claim 13 , further comprising at least one memory comprising a data table or knowledge database, comprising entries, one entry of said entries associating additional context information with at least one keyword.
15 . The platform according to claim 14 , wherein said at least one memory further comprises at least one data table relating to a type of the compliance anomaly, said at least one data table comprising examples of requests to obtain a correction recommendation that resulted in obtaining recommendations (to correct the compliance anomaly of said type, validated by the user, associated with a score, and the platform further comprises a reinforcement learning module configured to select at least one example of a request from said examples of requests to obtain a correction recommendation from a data table relating to said type of the compliance anomaly, as a function of said score, in response to a request to obtain additional context information received from said device.