Composer mic for DeepSeek Harness Web: tap-to-monitor live transcription and hold-to-talk, with host Edge TTS reply reading that streams while the model generates, echo-pause during reading, and tap-to-stop.
DSH plugin: text-only models (e.g. DeepSeek-V4) automatically see images via a vision model. Official surface-replace, cache-friendly, human transcript untouched. 纯文本模型自动识图桥
Archived-conversation manager for DeepSeek Harness: view (grouped by workspace), restore, and reveal transcript folders of archived conversations. Zero-build, copy-to-install plugin.
Hermes-style persistent memory + self-learning for DeepSeek Harness: MEMORY.md/USER.md stores injected as a system-prompt section refreshed every session, a single memory tool (add/replace/remove/atomic batch) with threat scanning, and a post-turn background review loop that replays the transcript through the llm service with a memory-only tool whi