Configuration & Usage

Configuration starts with profiles — each maps to a directory under ~/.dsh/profiles. Fill in your model API key and pick a workspace in Web UI's Settings → Models, then verify the merged result with dsh --dump-config.

This category covers configuring and using DeepSeek Harness Plugin (DSH plugin): dsh --profile usage, inspecting the merged config tree with --dump-config, setting models and selecting workspaces in the Web UI, and running one-shot tasks in headless mode. Most config changes apply immediately — verify as you go, and check dsh --dump-config if something looks off.

DSH-Plugin tutorial list

DSH configuration & usage FAQ

Where is DSH configuration stored?

Under the ~/.dsh data directory by default: each profile in ~/.dsh/profiles/<name> and session records in ~/.dsh/sessions. The path follows $DSH_HOME and is subject to the official docs.

How do --profile and dsh web relate?

Launcher arguments come first; --profile <name> selects the workspace. dsh web is shorthand for dsh --profile web, and headless maps to --profile headless.

How do I view DSH's current full config?

Run dsh --dump-config to see the merged config tree for the profile, or dsh --dump-default-config to compare defaults. Both only print config and change nothing.

How do I configure models in the Web UI?

Open Settings → Models, enter your API key and pick a model, save, then select a workspace and run tasks. Custom endpoints use OpenAI-compatible configuration per the official docs.

Do I need to restart DSH after changing config?

Most changes (like model routes) apply immediately; changes affecting dependencies or the profile structure are best followed by restarting dsh web so the environment reloads.