Dify Community Proposes Workflow Fix After Snippet Node IDs May Send Document Variables to Models Unresolved

On October 10, the Dify community reported a workflow variable parsing defect: in self-hosted Docker version 1.17.0, inserting document retrieval and LLM nodes through Snippet, then adding the retrieval result to a prompt with the variable selector, may cause the model to receive a raw placeholder and respond that it cannot read the content. The report includes reproduction steps and a minimal backend regular expression case. Issue report

The report attributes the cause to a format mismatch between the frontend and backend. The Snippet insertion code builds a new ID from the original node ID, a timestamp, and an index, with hyphens in between. The backend prompt parser’s allowed node ID characters, however, include only letters, numbers, and underscores. As a result, a document variable selected in the interface may still fail to match at runtime and pass through as literal prompt text. This also explains why similar workflows may behave differently depending on how their nodes were created. Reproduction and parsing path

Dify’s official documentation says workflow variables in prompts should be replaced with their actual values before being sent to the model. From an engineering perspective, this type of defect could be mistaken for a document retrieval failure or a limitation in model capability. When troubleshooting document question-answering workflows, teams should check not only upstream node outputs but also whether the prompt actually sent still contains variable markers. LLM node documentation

PR #43838, submitted the same day, instead generates node IDs using a numeric timestamp, incrementing it in case of collisions, while preserving remapping of graph references. The author says the added regression case failed before the fix and that 27 related tests passed afterward. They cover repeated insertions with the same timestamp, prompt references, and connection endpoints. However, no end-to-end testing in a deployment environment or validation with a real model was performed. Proposed fix

As of the time of review, the proposal remains open and awaiting review. Its scope is limited to preventing newly inserted nodes from getting incompatible IDs; existing saved workflows will not be fixed automatically. Engineering teams should track the merge and release status, and determine whether old workflows need migration. This case should not be generalized to mean that all Dify deployments are affected. Fix scope and status