ai-pi-or-elephant

Adversarial code review evaluates code for weaknesses by dispatching targets through Elephant Alpha via pi across multiple providers, returning a JSON score, summary, and structured issues to guide remediation.

To run a security review, pass a target file through the pi Elephant Alpha flow, which parses the target, dispatches the review, and returns a JSON score with identified issues and a synthesis of findings.

Yes, the workflow supports individual files, code changes, and security reviews across diverse codebases by parsing targets and coordinating multi-provider evaluation through Elephant Alpha via pi.

The review returns structured JSON findings containing a score, summary, and detailed issues, providing a synthesis of vulnerabilities found to guide your remediation efforts.

No external dependencies are required; the skill implements a deterministic workflow that handles target parsing, model dispatch to Elephant Alpha via pi, and output parsing internally.