The vulnerability exists in the FetchHttpResource function located in pkg/resolution/resolver/http/resolver.go. This function is responsible for fetching resources over HTTP for Tekton Pipelines. The core of the vulnerability is the use of io.ReadAll(resp.Body) without any limit on the size of the response body. This allows a malicious actor, with permissions to create TaskRuns or PipelineRuns, to specify a URL to a server they control. This server can then send an extremely large HTTP response. The Tekton resolver, upon receiving this response, attempts to read the entire body into memory, leading to excessive memory allocation. If the allocated memory exceeds the container's limit, Kubernetes will terminate the pod with an Out-of-Memory (OOM) error, causing a denial of service for all resolution services in the cluster. The fix, as seen in commit db61c71c6f4584e8b06e52d23b99b29d0bb078d5, introduces an io.LimitedReader to cap the number of bytes read from the response body, effectively mitigating the OOM threat.