MIT-licensed, local-first token/cost analytics dashboard that natively reads DeepSeek Harness session data, with heatmaps, cost tracking and quota monitoring.
DSH integration
Compatible
Author-claimed
Safety audit
Unaudited
Last verified
2026-08-29
License
MIT
01What can it help you accomplish?
Track and analyze token usage and cost across AI coding agents from one local dashboard
Heatmaps, exact token breakdowns, cost tracking and quota reset countdowns in a local web dashboard, plus an optional menu-bar companion
Developers and teams running DeepSeek Harness, Claude Code, Codex, Kimi and other agents who want local visibility into token and cost spend
Monitor subscription quota windows and reset countdowns
Quota tab with subscription window bars and reset timers, from local logs or opt-in live polling
Users on metered or subscription plans (Codex, Claude Code, Antigravity) who need to avoid hitting quota limits
02How to install into DeepSeek Harness
Prerequisites
- Python **3.10+**
- One or more [supported clients](docs/reference/SUPPORTED_CLIENTS.md) installed
Installation steps
- 01
pipx install tokdash
- 02
python3 -m pip install --user tokdash
- 03
tokdash setup
- 04
tokdash setup --auto --json
- 05
tokdash setup --dry-run
Verify the integration
- tokdash doctor
- tokdash doctor --json
Rollback
- tokdash uninstall
03DSH integration and capability boundaries
Reads DeepSeek Harness (dsh) sessions locally from $DSH_HOME/sessions — no API key, localhost-only, compatible with the DSH data layout
Exact token counts
agent session logs (local files)→Input / Output / Cache token breakdowns
Contribution calendar
historical usage records→2D heatmap + 3D isometric view with Tokens / Cost / Messages metrics
Quota tab
local session logs or opt-in live quota polling→subscription window bars with reset countdowns for Codex, Claude Code, and Antigravity
opt-in live polling calls each provider's own quota endpoint with your local CLI credentialsStatusline integration
local HTTP endpoint (Tokdash API)→live token-usage indicator for Claude Code's statusline or any agent that can hit a local endpoint
Companion Status Bar App
configured Tokdash endpoints→spend and subscription quota in the macOS menu bar or Windows notification area
the companion only contacts the Tokdash endpoints you configure
04Who is it for? When not to use it?
Good for
- Developers and teams running DeepSeek Harness, Claude Code, Codex, Kimi and other agents who want local visibility into token and cost spend
- Users on metered or subscription plans (Codex, Claude Code, Antigravity) who need to avoid hitting quota limits
Not for
- Optional quota polling uses your local CLI credentials only to call the provider's own quota endpoint and stores responses in the local usage SQLite DB; it is off by default and requires per-provider consent.
05Compatibility, maintenance and safety notes
- Tokdash is loopback-bound by default; the dashboard and local APIs are read-only and localhost-only, and fail silently if Tokdash isn't running.
- Optional quota polling uses your local CLI credentials only to call the provider's own quota endpoint and stores responses in the local usage SQLite DB; it is off by default and requires per-provider consent.
- Costs are computed from the bundled pricing database by default and may lag real provider pricing; treat them as estimates and verify against your billing source if it matters.
- macOS is supported; native Windows support is experimental.
MIT · actively maintained (latest release v2.4.0, 2026-08-28)
06Frequently asked questions
How does Tokdash connect to DeepSeek Harness?
It reads your DSH sessions locally from $DSH_HOME/sessions (default ~/.dsh) — no API key or network call needed. Each zstd log frame is decoded and forked-session prefixes are skipped so tokens are never double-counted.
Do I need to give Tokdash network access?
No. By default everything runs on localhost and reads local log files only. Network is used only if you opt into live quota polling, which calls each provider's own quota endpoint with your existing CLI credentials.
How do I install and verify Tokdash?
Install with `pipx install tokdash` (or `python3 -m pip install --user tokdash`), run `tokdash setup`, then `tokdash doctor` to confirm the runtime, service, port and data paths.
How accurate are the cost numbers?
Token counts come from what each client logs locally; costs are computed from the bundled pricing database and may lag real provider pricing. Treat them as estimates and verify against your billing source if it matters.
How do I remove Tokdash?
Run `tokdash uninstall` to reverse exactly what setup created; it keeps your usage history by default. Add `--purge` to also remove history.
07Related DSH workflows
dsh-cost-meter
by han-1413141
Cost tracking for DeepSeek Harness: session and model costs, token usage, budgets, provider balances, and coding-plan quotas. Bilingual English/Chinese UI.
dsh-usage-stats
by ychris12138
Provider balances, subscription quotas, and token-usage analytics for the DeepSeek Harness Web GUI (dsh web).
dsh-balance-meter
by ghost011118
DeepSeek account balance and session cost readout for the DeepSeek Harness Web GUI
dsh-plugin-langfuse
by linyp
Langfuse observability for DeepSeek Harness (dsh): exports agent sessions as OpenTelemetry trace trees (GenAI semconv) to Langfuse's OTLP endpoint
08Data and sources
DeepSeek Harness (`dsh`) usage and sessions are read locally from `$DSH_HOME/sessions/*/*/session.jsonl.zstd` (or the un…
| DeepSeek Harness | ✅ | ✅ |
This page is generated from the project’s public documentation, repository metadata and a structured parse of DSH Plugins; last verified on 2026-08-29. Found an error? Submit a correction.
Best DeepSeek Harness Plugins
Twelve plugins worth installing first — picked from the whole catalog, across every category.