dsh-date-wrapper
Run the following command in DeepSeek Harness:
dsh plugin install drscrewdriver/dsh-date-wrapper
Paste the following prompt into your AI chat to install this plugin:
Run dsh plugin install drscrewdriver/dsh-date-wrapper in the DeepSeek Harness terminal to install the plugin; source code is available at https://github.com/drscrewdriver/dsh-date-wrapper .
About this plugin
DSH's built-in @deepseek-ai/dsh-time-context injects roughly 280 characters of time metadata on every request, consuming tokens and writing per-turn events into the session log. dsh-date-wrapper replaces that with a single 46-character date line (~12 tokens) and lands it in DSH's existing runtime-context snapshot via systemPrompt.context. Because the platform deduplicates snapshots by text, no extra event is written while the date stays the same; the append-only surfaceOp keeps the prefix cache intact. In a measured 10-turn / 231-step session the legacy approach produced ~3.4 KB of events, while this plugin folds ~46 B into an already-present snapshot.
Activation of the plugin row is the on/off switch: when inactive, apply() never runs and no characters are injected. Configuration is limited to an IANA time-zone value; an invalid zone throws at startup rather than silently falling back to UTC. The plugin registers no settings namespace, reads no session data, makes no RPC call, imports nothing from @deepseek-ai/* packages, and has no browser-side component, staying within the systemPrompt.context contract stable from DSH 0.1.0-rc.7 through 0.1.3-alpha.2.
It suits DSH 0.1.x users in agent workflows who care about token cost and request size, need only the current date, and want the lightest possible replacement for the default time context. Do not enable it alongside @deepseek-ai/dsh-time-context, as the two overlap.
Use Cases
- Cut token overhead from time metadata on every request
- Replace the default time context with minimal footprint in agent workflows
- Inject the current date with automatic dedup to avoid repeated writes
Best For
- Agent workflow developers focused on request size and token cost
- DSH 0.1.x developers who need a leaner time context
- Prompt engineers seeking minimal-footprint plugin replacements
Related Plugins
A method pack that makes AI coding agents plan against your real baseline, prove completion with fresh evidence, and reduce reworks and unsafe changes.
Turns the DeepSeek Harness session into a captain that builds a durable sub-agent team, splits goals into dependency-aware tasks, and coordinates work via direct messages and a live Web UI.
Gives coding agents design judgment, letting Claude Code, Cursor, and 70+ agents generate and iterate high-quality UI, presentations, and graphics right from the terminal.
Run the Pi ecosystem's plugins on DeepSeek Harness, unmodified, via a compatibility layer that implements Pi's public extension ABI on DSH's native services.