🪨 why use many token when few token do trick — Claude Code skill that cuts 65% of tokens by talking like caveman
Telemetry and skills for Claude Code and Codex
Everr gives agents and builders direct access to runtime signals from local development, CI, and production. It uses a local collector, a GitHub Actions integration, and a hosted backend so the same data can be queried with SQL across all three places. It also bundles skills that help Claude Code, Codex, and Cursor work with telemetry.
Builders who want their agent to read test, CI, and runtime signals while they work.
You can ask for flaky failures, performance trends, and missing spans without opening dashboards or adding custom endpoints.
What it does
Local telemetry query
Runs a local collector with chDB so you can query OpenTelemetry data from your dev server, tests, or wrapped commands with SQL.
CI observability
Turns GitHub Actions workflow runs into traces with flakiness, performance, failure, and cost data.
Production telemetry routing
Sends the same OpenTelemetry data to standard OTLP collectors or to the Everr backend.
Bundled agent skills
Installs and updates skills for CI, telemetry setup, and telemetry usage in Claude Code, Codex, and Cursor.
Desktop and explorer views
Uses the same data in charts and a logs explorer in the desktop app.
README
Software delivery intelligence for developers and AI agents, local, CI, and production.
Website · Docs · Devlog · Contributing
Observability today is trapped behind dashboards, and most "AI integrations" are just legacy tooling with a ChatGPT wrapper slapped on top.
Everr gives you, and your AI agents, direct access to the signals that matter, wherever your code runs: on your laptop, in CI, and in production.
The problem
AI coding agents are making development faster than ever. Reading the codebase tells half the story; the other half is what the code actually does the moment it runs, and that half is usually trapped behind a dashboard the agent can't see.
By the time a regression shows up in a production graph, it's already too late. The bottleneck has moved from writing code to validating it.
How Everr helps
Think in Code. Everr exposes the same OpenTelemetry data, local, CI, and cloud, behind a read-only SQL surface. Agents already know how to write SQL, so they can answer "show me flaky failures on main last week" without us shipping a new endpoint for every shape of question. No screen-scraping, no log parsing, no custom RPC per workflow.
Built where the work actually happens. A local collector with an in-process ClickHouse, a CI integration that captures every workflow run, and a hosted backend that ties them together. Same primitives, same data, same answers, whether the agent is on your laptop, a sandbox in the cloud, or a teammate's machine three time zones away.
Editor- and agent-agnostic. VS Code, Cursor, Zed, JetBrains, Claude Code, Codex, Copilot, Everr doesn't replace your stack, it improves it. Bundled skills slot directly into the assistants you already use, so they reach for runtime data the moment a test fails or a span goes missing.
Key capabilities
- Local telemetry, queryable on the spot, A sidecar collector embeds chDB and exposes OpenTelemetry data from your dev server, tests, or any wrapped command over SQL.
everr local query "<SQL>"from the terminal; the same data drives charts and a logs explorer in the desktop app. - CI observability, Every GitHub Actions workflow run becomes a structured trace with flakiness scores, performance trends, failure patterns, and cost breakdowns. No YAML changes, no config files.
- Production observability, send the same OpenTelemetry data to any standard OTLP collector so it's available in your existing dashboards, or to local agents alongside everything else they already see.
- Bundled agent skills,
everr skills installdrops skills for working with CI, setting up telemetry, and using telemetry into Claude Code, Codex, or Cursor, globally or per project, kept in sync viaeverr skills update.
Get early access
Everr is currently in closed beta. Join the waitlist to get access.
- For CI: install the GitHub App on your repositories, no YAML changes, no config files, no modifications to your workflows.
- For local: install the CLI and run
everr local startto bring up the collector alongside your dev server, tests, or agent terminal.
Contributing
See CONTRIBUTING.md for development setup instructions.
License
This project is licensed under the GNU Affero General Public License, Version 3. Some components are subject to different license terms, see NOTICE for details.
Files in the repo
- .agents
- .changeset
- .claude
- .github
- .source
- .vscode
- clickhouse
- collector
- crates
- docs
- evals
- everr
- packages
- scripts
- todo
- .dockerignore
- .fallowrc.jsonc
- .gitignore
- .nvmrc
- AGENTS.md
- biome.json
- Cargo.lock
- Cargo.toml
- CHANGELOG.md
- CLAUDE.md
- CONTEXT.md
- CONTRIBUTING.md
- docker-compose.yaml
- lefthook.yml
- LICENSE
- Makefile
- NOTICE
- package.json
- pnpm-lock.yaml
- pnpm-workspace.yaml
- README.md
- skills-lock.json
- turbo.json
- vitest.config.ts
Discussion (0)
Ask about usage, or say what you built with itSign in to join the discussion.
No comments yet. Be the first to say what this is good for.
More tools
The best-benchmarked open-source AI memory system. And it's free.
Orca is the ADE for working with a fleet of parallel agents. Run any coding agent with your own subscription. Available on desktop, mobile and remote runtime.

A cross-platform desktop All-in-One assistant for Claude Code, Codex, OpenCode, OpenClaw, Grok Build & Hermes Agent. Only official website: ccswitch.io
Never stop coding. Free MIT AI gateway: one endpoint, 352 providers (150+ free), 1200+ models Kimi, Claude, GPT, Gemini, GLM, DeepSeek, MiniMax. Works with Claude Code, Codex, Cursor, OpenCode, Cline & Copilot. Quota-aware auto-fallback, RTK+Caveman compression saves 15-95% tokens, MCP/A2A, Desktop/PWA. Built by 550+ contributors
Compress tool outputs, logs, files, and RAG chunks before they reach the LLM. 20% fewer tokens for coding agents, 60-95% fewer tokens for JSON, same answers. Library, proxy, MCP server.