> ## Documentation Index
> Fetch the complete documentation index at: https://costhq.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.
# Introduction to CostHQ
> Discover how CostHQ enables transparent AI agent cost tracking and LLM spend monitoring for developers and teams using Claude Code, OpenAI, or local models.
AI agents are powerful — but they burn tokens fast, and most developers have no idea what a single coding session actually costs until the invoice arrives. CostHQ solves that by giving you a lightweight CLI and Node.js SDK that wraps your existing workflow, captures every meaningful signal (time, files, commits, and AI spend), and stores everything locally in a SQLite database. You get a full picture of each session without changing how you work or sending any data off your machine.
## Four things CostHQ tracks
Every session CostHQ records covers four dimensions of cost and activity:
* **Time** — Wall-clock duration from session start to end, measured to the second and displayed in a human-readable format (e.g. `14m`).
* **Files** — The number of files modified during the session, detected automatically from your working directory.
* **Commits** — Git commits made while the session is active, so you can correlate code output with AI spend.
* **AI Cost** — Token usage logged against 42+ built-in model prices from Anthropic, OpenAI, Codex, DeepSeek, Azure, Cohere, Google, and more.
## How it works
CostHQ gives you four ways to track a session, so you can pick the method that fits your workflow without any forced migration.
Use `cs run ` to start a session, launch a proxy, run your command, and print a cost summary — all in one step.
Run `cs proxy start` to spin up a local HTTP proxy on port 3739 that transparently intercepts Anthropic and OpenAI API calls and logs token usage automatically.
Use `cs start` to open a session and `cs end` to close it, with full control over naming and notes. Ideal for scripting or multi-step workflows.
Install the CostHQ skill via `clawhub install costhq` to add native session tracking inside Claude Code and other OpenClaw-compatible agents.
## Who it's for
CostHQ is designed for three groups of people who all feel the same pain — invisible AI spend.
**Individual developers** running Claude Code, Copilot, or custom agents who want a simple answer to "how much did that task cost me today?"
**Teams** that need shared visibility into AI spend across members, with budget alerts and a centralized dashboard to prevent runaway costs on long-running agent jobs.
**Agent framework authors** who want to instrument their SDKs and give end users first-class cost reporting without building infrastructure from scratch.
## Privacy & data
CostHQ is built on a local-first philosophy. Your session data never leaves your machine unless you explicitly opt in to a paid plan with cloud sync.
All session data is stored in a local SQLite database at `~/.costhq/sessions.db`. CostHQ collects no telemetry and makes no outbound network requests on the free tier.
The database runs in WAL mode for safe concurrent access and stays out of your project directories entirely — nothing is written to your repo or `.gitignore`.
## Pricing overview
CostHQ's core CLI is free forever. Paid tiers add collaboration and cloud features for teams that need them.
| Tier | Price | Highlights |
| -------------- | ------------ | ---------------------------------------------------------------------------------------------- |
| **Free** | \$0 | Local CLI, SQLite storage, basic dashboard |
| **Pro** | \$12/mo | Cloud sync, semantic caching proxy, shareable stats card |
| **Team** | \$25/seat/mo | Shared budget wallets, centralized dashboard, Slack/Teams alerts, GitHub Actions CI/CD breaker |
| **Enterprise** | Custom | SAML SSO, SOC2 compliance, on-premise proxy |