AI Agent Hub
Back to plugins
dsh-balance-tasks preview

dsh-balance-tasks

Client Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install Bil812/dsh-balance-tasks

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install Bil812/dsh-balance-tasks in your terminal to install this plugin from GitHub (https://github.com/Bil812/dsh-balance-tasks), then restart dsh web.

About this plugin

When batching tasks on DeepSeek with a CNY balance, the most pressing question is how many more runs you can actually get before the money runs out. dsh-balance-tasks answers it with a small floating card in the bottom-right corner that shows your live balance, an estimated number of remaining tasks priced from the current context onward, and a countdown to the next peak or off-peak tier switch. One click refreshes everything without leaving your workspace.

The pricing model mirrors dsh-conversation-cost exactly: miss-rate for uncached input, hit-rate for cached reads, and out-rate for output tokens; peak pricing during peak hours and half-price during off-peak. The remaining-task estimate starts from your current context and accumulates task-by-task, folding the historical average consumption and the existing context into a single formula so you never see two contradictory numbers. The ledger is persisted with atomic writes and survives restarts.

Built for researchers and engineers who burn CNY balances on multi-agent, long-context batch workloads on DeepSeek, and who want a single glance to decide whether todays budget holds or when to shift to off-peak hours for maximum value.

Screenshots

Use Cases

  • Check remaining task budget before kicking off a batch run
  • See the current peak or off-peak tier and the countdown to the next switch
  • Compare per-model remaining tasks and average cost across parallel workloads

Best For

  • Researchers running multi-agent, long-context batch jobs on a CNY balance
  • Engineers who need precise control over DeepSeek spend and usage
  • Cost-optimization teams that track peak vs off-peak pricing for maximum value