yflmq001/dsh-cost-tracker
Token cost tracking for DeepSeek Harness
Token cost tracking for DeepSeek Harness with per-model configurable pricing and peak/off-peak rates: prices every finalized LLM call against a per-model pricing table you configure (cache-hit/cache-miss input, output, optional peak-window rates), publishes a per-session cost projection (total/peak/off-peak/per-model) consumed by the Web UI as a live session cost readout, and flags calls in a configured peak window. Two billing-correctness choices: cache writes are billed at the miss rate (not the hit rate), and unknown models are never silently priced — they surface as 'unconfigured' with a placeholder.
Install
dsh plugin --profile web add github:yflmq001/dsh-cost-trackerGitHub install per README (English): dsh plugin --profile web add github:yflmq001/dsh-cost-tracker — the plugin ships a dsh.bundle layer, so dsh plugin add enables it automatically. Pricing starts empty (models: {}); fill in your models under the plugin's config. To override defaults, add a cost-tracker row to the profile's cordis.patch.yml — address the row by its id (cost-tracker), not name, or the override is silently dropped. npm package dsh-cost-tracker 1.0.1 exists on the registry (verified 2026-08-28); README documents the github: path.
Compatibility
DeepSeek Harness web profile. Per-model configurable pricing with peak/off-peak rates; cross-session bill persists through dsh's storage domain.
Details
- Repo: yflmq001/dsh-cost-tracker
- Category: Other
- Stars: 3
- Version: GitHub source per README (npm dsh-cost-tracker 1.0.1 on registry, verified 2026-08-28)
- Last push: 2026-08-26
- First seen: 2026-08-14
Recent updates
per-model pricing table; peak/off-peak rates; live per-session cost readout; cache writes at miss rate; unknown models stay unconfigured.
FAQ
- How are costs priced?
- Every finalized LLM call is priced against a per-model table you configure — cache-hit/cache-miss input, output, and optional peak-window rates.
- Are cache writes priced correctly?
- Yes — cache writes are billed at the miss rate (providers charge writes at the miss tier), not the hit rate.
- What about models without a price?
- Unknown models are never silently priced: they surface as 'unconfigured' with a placeholder until you add the price.
Alternatives
dongsheng123132/dsh-cost · YYTbit/dsh-plugin-cost-tracker · GiantGKL/dsh-cost