Description

A vulnerability has been found in HKUDS AI-Trader up to d03ff6c056b32ced735adf7c19ed8175adb1c8df. The affected element is an unknown function of the file service/server/routes_agent.py of the component selfRegister API Endpoint. Such manipulation of the argument initial_balance leads to business logic errors. The attack may be launched remotely. This attack is characterized by high complexity. The exploitability is described as difficult. The exploit has been disclosed to the public and may be used. This product operates on a rolling release basis, ensuring continuous delivery. Consequently, there are no version details for either affected or updated releases. profit_percent_for_display() divides by INITIAL_CAPITAL + deposited, and challenge scoring's return_pct also normalises against the attacker-inflated starting_cash. So an inflated initial_balance does not yield artificial percent returns - it inflates the absolute cash/equity column only, which is a cosmetic/leaderboard-gaming concern in a simulated game.

Executive Summary

Technical Details

Mitigations

References

Additional Metadata

{
  "score": "3.7",
  "pubDate": "2026-09-03T13:06:19.590Z",
  "pubdate": "2026-09-03T13:06:19.590Z",
  "executiveSummary": "A business logic vulnerability exists within the HKUDS AI-Trader selfRegister API endpoint, specifically affecting the handling of the initial_balance argument within service/server/routes_agent.py.\nThis flaw allows remote attackers to manipulate the initial capital allocated to a trading agent, leading to an inflation of absolute cash and equity metrics reported within the simulated trading environment.\nThe vulnerability is characterized by its high complexity and difficulty of exploitation. While the affected software operates on a rolling release cycle without explicit version numbering, all builds up to commit d03ff6c056b32ced735adf7c19ed8175adb1c8df are confirmed susceptible.\nThe impact is primarily cosmetic and related to leaderboard manipulation or simulated game performance, as internal logic for calculating percentage-based returns utilizes normalization against the provided initial_balance, preventing direct financial exploitation of return percentages.\nThe exposure is network-based, allowing remote interaction with the selfRegister API endpoint.",
  "technicalDetails": "The root cause of this vulnerability lies in the insufficient server-side validation of the initial_balance parameter submitted to the selfRegister API endpoint located in service/server/routes_agent.py.\nThe application accepts a user-supplied numeric value for the agent's starting capital without enforcing strict boundaries or sanity checks against expected simulation parameters.\nExploitation is achieved by transmitting a crafted request to the selfRegister endpoint containing an arbitrarily large value for the initial_balance field. Upon successful processing, the backend service registers the agent with the attacker-specified capital amount.\nThe technical impact on data integrity is manifested in the reporting functions of the trading platform. Specifically, the function profit_percent_for_display() performs division operations using the sum of INITIAL_CAPITAL and the deposited amount; similarly, the challenge scoring mechanism's return_pct calculation normalizes performance metrics against the inflated starting_cash provided during registration.\nBecause these functions normalize performance against the provided initial_balance, the attacker cannot generate fraudulent percentage-based returns that would trick the system's performance algorithms. However, the logic failure results in the absolute cash and equity columns of the leaderboard or user dashboard reflecting the inflated starting balance.\nThis behavior constitutes a business logic error where the integrity of the simulated competitive environment is compromised, allowing users to artificially inflate their absolute standing in the application's reporting modules. The attack is considered high complexity due to the requirement of understanding the specific interaction between the registration input and the downstream data processing functions within the HKUDS AI-Trader architecture."
}