@icrefin

Workspace-level MCP server manager for DeepSeek Harness: one profile-scoped server registry, per-workspace selection, managed from a sidebar panel.

Profile-level skill manager for DeepSeek Harness: a flat skill inventory with enable/disable, discovery diagnostics and a restorable trash, in a sidebar panel.

Names the provider beside the model in DeepSeek Harness's composer control (Provider/Model) while keeping the provider-grouped picker.

Per-model reasoning effort levels inside DeepSeek Harness Settings -> Models, written to that model's reasoningEfforts in cordis.patch.yml.

Edit the AGENTS.md instruction files DeepSeek Harness actually loads - user-global and per-workspace - from a left-sidebar panel in the GUI.

Push DeepSeek Harness agent notifications to your phone and watch via ntfy, Bark, Gotify, Telegram or a webhook.

Keeps the DeepSeek Harness context-usage ring honest: an empty usage sample during a retry no longer collapses it to 0%, and its percentage matches the figures beside it.

Max-retries control for DeepSeek Harness: one Settings -> Model card that sets how many times a failed model request is retried.