# US 2026/0227969 A1 · Context-aware code generation with pre-fetching and automatic prompt distillation > Agents: this data is available as JSON from the patentgrep API and MCP server — see https://patentgrep.com/llms.txt - Number: US 2026/0227969 A1 - Document ID: US20260227969A1 - Status: Published application - Kind: A1 - Application number: 19/465,542 - Filed: 2026-01-30 - Published: 2026-08-06 - Earliest claimed: 2025-02-06 - Assignees: For Good AI Inc. - Applicants: For Good AI Inc. - Inventors: Andrey Filev; Dmitrii Krasnov - Web page: https://patentgrep.com/patent/US20260227969A1 ## CPC - G06F 8/31: {Programming languages or programming paradigms} - G06F 8/35: model driven - G06F 8/427: {Parsing} - G06N 3/0475: Generative networks - G06N 5/022: Knowledge engineering; Knowledge acquisition - G06N 20/00: Machine learning - G06F 8/33: Intelligent editors - G06F: ELECTRIC DIGITAL DATA PROCESSING (computer systems based on specific computational models G06N) - G06N: COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS ## Abstract The present disclosure relates to systems and methods for context-aware code generation. A method for pre-fetching context includes identifying a user's coding project in an integrated development environment, retrieving context information from project management tools, task tracking systems, documentation platforms, or web browser data prior to receiving a user request, parsing the retrieved context using natural language processing, generating and storing contextual data, and generating a response based on the stored contextual data. A system for gathering background context monitors user activity within a development environment, gathers context through explicit user input, implicit inference, or industry-standard datasets, processes the context to generate structured contextual data, and generates code suggestions based on the stored data. A method for automatic prompt distillation monitors code written by a user, analyzes the code in real-time, and generates new prompts based on the analyzed code. ## Claims ### Claim 1 A method comprising: identifying a user's coding project in a development environment; prior to receiving a request from the user in the development environment: retrieving context information from one or more sources comprising at least one of project management tools, task tracking systems, documentation platforms, or web browser data; parsing the retrieved context information using natural language processing techniques to identify at least one of keywords, concepts, or relationships relevant to a coding task; generating contextual data associated with the user's coding project based on the parsing of the retrieved context information; and storing the contextual data in a memory; receiving the request from the user in the development environment; and generating a response to the request based on the contextual data stored in the memory. ### Claim 2 (depends on 1) The method of claim 1, wherein retrieving context information comprises accessing at least one of JIRA tickets, Asana tasks, or Notion pages associated with the user's current or past work. ### Claim 3 (depends on 1) The method of claim 1, wherein parsing the retrieved context information further comprises extracting key information comprising at least one of project requirements, specific feature descriptions, or technical constraints from work items. ### Claim 4 (depends on 1) The method of claim 1, further comprising: detecting that the user has switched to a different work item or that a work item has been updated; retrieving updated context information based on the detected switch or update; and updating the stored contextual data based on the updated context information, wherein retrieving the updated context information is performed at predetermined intervals. ### Claim 5 (depends on 1) The method of claim 1, wherein the development environment comprises one of an integrated development environment (IDE) or a cloud-based coding agent. ### Claim 6 (depends on 1) The method of claim 1, further comprising: receiving a manual trigger from the user to initiate or refine the context pre-fetching; and in response to the manual trigger, retrieving additional context information and updating the stored contextual data. ### Claim 7 (depends on 1) The method of claim 1, wherein generating the response comprises initializing a code generation model with the stored contextual data or providing the stored contextual data as additional input during a code generation process. ### Claim 8 (depends on 1) The method of claim 1, wherein when the user is working on a work item related to implementing a new feature, retrieving context information comprises: identifying the work item associated with the new feature; retrieving specific requirements, security protocols, and relevant API documentation associated with the new feature; and incorporating the retrieved specific requirements, security protocols, and relevant API documentation into the stored contextual data. ### Claim 9 A system comprising: one or more processors; and non-transitory computer-readable media storing instructions that, when executed by the one or more processors, cause the one or more processors to perform operations comprising: monitoring activity of a user within a development environment; gathering background context through at least one of explicit user input, implicit inference from the activity of the user, or utilization of industry-standard datasets, wherein the background context is used to enhance code completion and generation tasks; processing the background context to generate structured contextual data; storing the structured contextual data in a memory; receiving a code generation request from the user; and generating code suggestions in response to the code generation request based on the stored structured contextual data. ### Claim 10 (depends on 9) The system of claim 9, the operations further comprising: receiving explicit input from the user comprising relevant information about at least one of a coding task, project requirements, or specific technical details; parsing the explicit input to extract the relevant information; and incorporating the extracted relevant information into the stored structured contextual data. ### Claim 11 (depends on 9) The system of claim 9, the operations further comprising: analyzing user actions within the development environment comprising monitoring at least one of the user's coding patterns, file access history, or interactions with documentation; inferring implicit context from the analyzed user actions; and incorporating the inferred implicit context into the stored structured contextual data. ### Claim 12 (depends on 9) The system of claim 9, the operations further comprising: recording outcomes of previous code generation attempts comprising at least one of successful paths or failed paths; extracting guidance data from the recorded outcomes indicating approaches to pursue or approaches to avoid; and incorporating the guidance data into the stored structured contextual data for use in subsequent code generation tasks. ### Claim 13 (depends on 9) The system of claim 9, the operations further comprising: retrieving the stored structured contextual data from the memory; initializing an underlying language model used for code generation with the retrieved stored structured contextual data; and fine-tuning the underlying language model based on the retrieved stored structured contextual data. ### Claim 14 (depends on 13) The system of claim 13, the operations further comprising: determining at least one of a type of algorithm or a programming paradigm the user is working on based on the stored structured contextual data; filtering or prioritize code suggestions based on the determined type of algorithm or programming paradigm; and outputting the filtered or prioritized code suggestions to the user. ### Claim 15 (depends on 9) The system of claim 9, the operations further comprising: detecting at least one of changing requirements or shifts in the user's focus based on the activity of the user; gathering updated background context in response to the detected changing requirements or shifts; and updating the stored structured contextual data based on the gathered updated background context. ### Claim 16 A method comprising: receiving an initial prompt from a user for a coding task; executing a code generation agent based on the initial prompt; monitoring results of an execution of the code generation agent in conjunction with a development environment; analyzing the results, wherein the analyzing comprises at least one of: evaluating code generated during the code generation agent run; evaluating execution outcomes comprising at least one of compilation results, test results, or runtime behavior; or evaluating observations of behavior of applications associated with the development environment; generating new prompts based at least in part on analyzing the results, wherein the new prompts are generated prior to receiving additional natural language input from the user; storing the new prompts in a memory; providing the new prompts to the user via a chat interface associated with the integrated development environment; and generating code suggestions or responses based on the new prompts. ### Claim 17 (depends on 16) The method of claim 16, wherein analyzing the results further comprises: extracting relevant information from the code generated during the code generation agent run comprising at least one of function names, variable declarations, or code comments; storing the extracted relevant information; and generating the new prompts based on the stored extracted relevant information. ### Claim 18 (depends on 17) The method of claim 17, further comprising: applying natural language processing techniques to the stored extracted relevant information; converting the stored extracted relevant information into coherent prompts formulated as at least one of questions or statements; and using the coherent prompts to guide a subsequent code generation process. ### Claim 19 (depends on 16) The method of claim 16, wherein analyzing the results further comprises: accessing an entire codebase associated with the user's coding project; analyzing at least one of related files, imported libraries, or project structure from the accessed codebase; generating codebase context data based on the analysis; and incorporating the codebase context data when generating the new prompts. ### Claim 20 (depends on 16) The method of claim 16, further comprising: retrieving pre-fetched context comprising at least one of work item information from a project management tool or task tracking system; combining the pre-fetched context with the analyzed results to infer a most logical next step for the user; generating the new prompts based on the inferred most logical next step; and presenting the new prompts to the user via the chat interface or automatically executing a subsequent code generation task based on the generated new prompts.