Test Agent
An agent system that uses AI to automatically generate test cases, execute tests and produce reports. It automates the full workflow, including interface import, constraint identification, test case generation, assertions and database validation.
Timeline (last 90 days)
The single-interface demo is not enough to prove performance under complex business flows, and the accuracy of the constraint model also depends on the quality of the first manual confirmation.
A developer spent more than four months building a testing agent.
The demo of this testing agent started from a completely wiped system, with all data tables cleared and no projects, environments or interfaces.
The system's interface import supports two methods; in the demo, reading a local message file for parsing was chosen, and after uploading the message file the system parsed out two interfaces.
In the demo, the author selected the homepage interface and sent the request directly; the status code returned 200 and the response data was normal.
When designing test cases for the first time, the system automatically identifies the Chinese meaning, data type and value range of each parameter, and requires one round of manual confirmation.
In the demo, the system identified that an enum field has only two values, 1 and 2, the status field is limited to the range 1 to 12, the password length is limited to 6 to 8 characters, and the phone number field is marked as needing to match the real number format.
The phone numbers generated by the system are real numbers that conform to the specification, not random digits; values for fields such as usernames fall strictly within the preset ranges.
The test cases generated by the system cover testing methods such as orthogonal combinations, boundary values, missing options and missing required fields, and each case is tagged with a priority.
Each generated test case contains five layers of assertions, including validation of the outermost status code, assertions on the response body and assertions on the database.
8 more timeline events
All Facts (18)
The single-interface demo is not enough to prove performance under complex business flows, and the accuracy of the constraint model also depends on the quality of the first manual confirmation.
50%UnverifiedA developer spent more than four months building a testing agent.
50%UnverifiedThe phone numbers generated by the system are real numbers that conform to the specification, not random digits; values for fields such as usernames fall strictly within the preset ranges.
50%UnverifiedThe test cases generated by the system cover testing methods such as orthogonal combinations, boundary values, missing options and missing required fields, and each case is tagged with a priority.
50%UnverifiedEach generated test case contains five layers of assertions, including validation of the outermost status code, assertions on the response body and assertions on the database.
50%UnverifiedFor business constraints such as username uniqueness, the system automatically generates delete operations to ensure data independence between test cases.
50%UnverifiedThe generated test cases require no manual modification of request parameters, priorities, specific values or assertions, and are ready to execute out of the box.
50%UnverifiedThe author compared it with similar tools on the market: previous tools often still required manual editing of request parameters, priorities and assertions after generating test cases, and this system's difference lies in automating these steps as much as possible.
50%UnverifiedThe constraint model only needs to be confirmed once during the first design, and subsequent test case generation for similar interfaces reuses it directly without popping up confirmation again.
50%UnverifiedThe system's interface import supports two methods; in the demo, reading a local message file for parsing was chosen, and after uploading the message file the system parsed out two interfaces.
50%UnverifiedWhen designing test cases for the first time, the system automatically identifies the Chinese meaning, data type and value range of each parameter, and requires one round of manual confirmation.
50%UnverifiedWeb/APP automation agents are not true testing agents; a true testing agent must be able to distinguish between test cases, test steps, test assertions, and test reports, and requires runtime context and execution context.
50%UnverifiedA failed case appeared in the demo because the parameter constraint model was incorrectly set to "user already exists," causing the generated request parameters to mismatch expectations.
50%UnverifiedIn the demo, the author selected the homepage interface and sent the request directly; the status code returned 200 and the response data was normal.
50%UnverifiedThe demo of this testing agent started from a completely wiped system, with all data tables cleared and no projects, environments or interfaces.
50%UnverifiedIn the demo, the system identified that an enum field has only two values, 1 and 2, the status field is limited to the range 1 to 12, the password length is limited to 6 to 8 characters, and the phone number field is marked as needing to match the real number format.
50%UnverifiedIn the demo, 32 test cases were generated and reviewed in batches of 10 per group; the review results fell into three categories: 9 cases that passed directly and could be committed, defective cases, and cases requiring manual confirmation.
50%UnverifiedAfter the review passed, the 9 cases were committed directly, no test case was manually modified at any point in the whole process, and the tests were then run directly.
50%