Check code against user intent with review agents and tests
All good ideas, I have been doing something similar for half a year. Task: All good ideas, I have been doing something similar for half a year. The most important thing is to reduce bugs and ensure the code does what you wanted. Writing code tests, logging, interacting with the running app are essential. For intent realignment I like to have the agent re-read all user messages posted in that project. It only takes a few minutes even on a 10,000 message project. Then use this fresh derivation of user intent to check code and tests. For long spans of work I prefer to generate a todo list which remains as a document in the project. Each closed gate tracks feedback from the agent on what happened. This task file is also reviewed twice - once before implementation and once after. So the formula is - review agents - tests - user message logs.; tools: custom-built or unnamed agent. Human role: The original post does not say how a person stayed involved.; limitations: Prerequisites include access to full project message logs and a codebase with tests. The source does not specify costs or integration constraints. Human approval is required for the two-stage review of the task file. Source reliability is secondary, based on a single user comment..
Tools
custom-built or unnamed agent
Published on Agent Work Atlas:
Overview
All good ideas, I have been doing something similar for half a year.
Workflow
All good ideas, I have been doing something similar for half a year. The most important thing is to reduce bugs and ensure the code does what you wanted. Writing code tests, logging, interacting with the running app are essential. For intent realignment I like to have the agent re-read all user messages posted in that project. It only takes a few minutes even on a 10,000 message project. Then use this fresh derivation of user intent to check code and tests. For long spans of work I prefer to generate a todo list which remains as a document in the project. Each closed gate tracks feedback from the agent on what happened. This task file is also reviewed twice - once before implementation and once after. So the formula is - review agents - tests - user message logs.
Reported result
The source does not report measured outcomes or success metrics. It only states the author has used this approach for half a year to reduce bugs and ensure code matches intent. A proposed acceptance check is to verify that the agent's intent derivation matches the final code behavior in a sample project.
Human role
The original post does not say how a person stayed involved.
Limitations
Prerequisites include access to full project message logs and a codebase with tests. The source does not specify costs or integration constraints. Human approval is required for the two-stage review of the task file. Source reliability is secondary, based on a single user comment.
Try it
Act as a code review agent. I will provide [replaceable inputs: project message logs, current codebase, and test suite]. First, re-read all user messages to derive the current user intent. Second, generate a todo list document tracking this intent. Third, check the existing code and tests against this derived intent. Fourth, record feedback for each closed gate in the todo list. Output a report showing the intent derivation, code/test mismatches, and the updated todo list. Before finalizing, perform a dry run on a small sample of messages and code to confirm the intent extraction is accurate. Wait for my confirmation before processing the full project. If inputs are missing, ask for the specific message logs or code files needed.
Source reports are not independently verified. Publication date is the catalog publication date, not the original article date.