Caprock — Development skill for Claude Code
Local dashboard for every Claude Code session on your machine — live activity, cost per repo, loop alerts.
How to install Caprock
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open dspv/caprock and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Caprock does
Local dashboard for every Claude Code session on your machine — live activity, cost per repo, loop alerts. No telemetry.
Alternatives in Development
- Claude Usage — A local dashboard for tracking your Claude Code token usage, costs, and session history 864 ★
- Dario — A local OpenAI- and Anthropic-compatible proxy that brings your Claude Pro/Max subscription to Cursor, Cline 485 ★
- Claude View — 10 Claude sessions running 104 ★
README
Caprock
See what your Claude Code is actually doing.

*Real capture, 46 seconds. One session working; the bars advance as the minutes roll over and the cost ticks with them.*
[](https://github.com/dspv/caprock/releases) [](LICENSE) [](https://github.com/dspv/caprock/actions/workflows/ci.yml) 
# macOS / Linux
brew install dspv/tap/caprock
# Windows
scoop bucket add dspv https://github.com/dspv/scoop-bucket
scoop install caprock
caprock up # opens localhost:4173; offers to set up hooks + plan-limit status line
Have Go? `go install github.com/dspv/caprock/cmd/caprock@latest` — the dashboard is embedded, so it works with no Node build. No package manager? Grab a binary from [Releases](https://github.com/dspv/caprock/releases).
Or let Claude Code install it — paste one prompt into a fresh Claude Code message.It picks the right command for your machine, so there is nothing to choose. The `--yes` is load-bearing: consent for the hook is refused rather than assumed when stdin is not a terminal, which is what an agent's shell gives it, so without the flag you would get a dashboard that never sees a session.
The wording was tested against five agents given no other context. An earlier draft was refused outright by one of them, and the two lines it objected to were both ours meaning well: "do not start or type into any claude session" (our Rule 7, which reads as *do not tell the user*) and "show me the error instead of trying another source" (meant to stop a lookalike package being installed, which reads as *stop verify
Related Skills
Ullage
Context-window telemetry for Claude Code and OpenAI Codex, plus Cursor agent activity: a macOS menu bar item s
Cctally
Claude Code usage tracker and local dashboard for Pro/Max subscription limits - weekly cost-per-percent trend,
Agenthropic
Self-hosted, local-first dashboard for observing Claude Code agent and subagent activity — a persisted subagen
Argus Code
See what Claude Code actually did. Local-first observability for every session: which tools ran, what each sub
Claude Code Agent View Demo
Web companion dashboard for Claude Code Agent View — live status, telemetry, and per-session post-mortem
Fleet Dashboard
A live board of every coding-agent session on your Mac — Claude Code, Codex, Antigravity — what each is doing
Related Agents
Community Dashboard Curator
Community telemetry and dashboard aggregator. Scans and compiles metrics across Skool cohorts, Discord activit
Dashboard Support
Support agent for the Agent Dashboard - guided install and setup diagnostics on the local machine, plus remote
Ollama Agent
Runs a local model served by Ollama on this machine. One mode - consult (no machine access; the endpoint is a