Adding a new detector

Slither's plugin architecture lets you integrate new detectors that run from the command line.

Detector Skeleton

The skeleton for a detector is:

from slither.detectors.abstract_detector import AbstractDetector, DetectorClassification


class Skeleton(AbstractDetector):
    """
    Documentation
    """

    ARGUMENT = 'mydetector' # slither will launch the detector with slither.py --mydetector
    HELP = 'Help printed by slither'
    IMPACT = DetectorClassification.HIGH
    CONFIDENCE = DetectorClassification.HIGH

    def detect(self):
        return []

• ARGUMENT lets you run the detector from the command line
• HELP is the information printed from the command line
• IMPACT indicates the impact of the issue. Allowed values are:
  • DetectorClassification.INFORMATIONAL: printed in green
  • DetectorClassification.LOW: printed in green
  • DetectorClassification.MEDIUM: printed in yellow
  • DetectorClassification.HIGH: printed in red
• CONFIDENCE indicates your confidence in the analysis. Allowed values are:
  • DetectorClassification.LOW
  • DetectorClassification.MEDIUM
  • DetectorClassification.HIGH

detect() needs to return a list of findings. To facilitate the automation of Slither, a finding is a dictionary containing a vuln key associated with the vulnerability name and additional information according to the vulnerability itself.

An AbstractDetector object has the slither attribute, which returns the current Slither object, and the log(str) function to print the result.

Integration

You can either integrate your detector into slither by:

• Adding it in slither/detectors and its instance in main.py#L92-L108
• Or by creating a plugin package (see the skeleton example).

Test the detector

Create a unit-test per detector in tests and update scripts/travis_test.sh to run the unit-test automatically.

Example

backdoor.py will detect any function with backdoor in its name.