Overall Score
In order to provide a simple framework to allow our users to evaluate tokens quickly and accurately, we designed a novel scoring system which aids in visually identifying how secure tokens are:
Compromised
A score of 0, dangerous and should be avoided.
This score is assigned for tokens that are already compromised. Meaning interactions with this token contract will almost certainly result in a loss of some or all funds. Examples of this, are tokens that are known honeypots.
Heavily Compromised
A score between 0 and 25, highly risky to interact with.
This score is assigned for tokens that are highly compromised to the extent that interacting with these tokens will likely result in a loss of funds. If any of the four component scores are extremely low, this will result in the token having a score that is heavily compromised.
Major Vulnerabilities
A score between 25 and 50, these tokens have some risk.
This score is assigned for tokens that have major vulnerabilities, which may result in loss of funds if the token is changed or modified in some way. Interacting with a contract with this score could be safe, but depending on the type of vulnerability, we would recommend caution when interacting with contracts with a major vulnerability score.
Moderate Vulnerabilities
A score between 50 and 75, these tokens are generally fine to interact with.
This score is assigned for tokens that are generally safe, but may have aspects or elements that lower their score, such as a bad holder profile, or low (but not unusual levels of) taxes.
Generally Safe
A score between 75 and 100, these tokens are generally safe.
This score is assigned for tokens that our Cognitive Framework thinks are generally safe. This means there are no clear ways of exploiting the token, and the token is likely safe to interact with. Tokens with this classification also have a non-malicious holder profile, based on the Cluster Maps liquidty report of holder interactions.
Methodology
The scoring system is calculated by our Cognitive Framework, which determines all the relevant factors that go into making a token safe. This, while highly effective, is not yet a completely error free way of determining how safe tokens are. This is due to the degree of variance found in analysing anything with AI-based models. However, these scores are a highly effective method of determining safety, with few false positive outputs. In plain terms, this means when low scores are output, this is likely because there is a large degree of risk in interacting with the contract.
Self Improvement
The scoring system itself is adaptive, and self-improving meaning it has the ability to learn with an ever-increasing datasets of new tokens, new rugs, and new contract features. Our scoring system distills a wealth of data from different aspects of the token to provide a comprehensive, in-depth evaluation that goes beyond any deterministic or surface level metrics that currently exist.
Last updated
