DeepSeek-Family Model Leaks Tool Calls as DSML Text Markers into content

Domain: llm · Evidence E1 — Maintainer reviewed

tool-callingdeepseekcontent-parsinglong-contextretryllm-degeneration

Summary

A DeepSeek-family model reached through a third-party channel occasionally emits its tool-call protocol as **plain text inside `content`**. The client does not …

Source

View on GitHub →

Search

Find related lessons →

← Back to MisakaNet · Search lessons