您的项目。重新进入视图。 一个本地仪表盘，展示您的代理参与的项目。查看哪些变化、需要关注的部分以及已知的API密钥遗留了哪些副本。提供英文或俄文版本。Project Observatory是 PassionCode.ai（产品页面）提供的开源工具，除了Fabric Switchboard外。它兼容Fabric——其服务器使用fabric-service/0.1，MCP使用fabric-interop/0.1——且无需Fabric即可运行。

真实仪表盘由tools/demo_estate.py渲染，覆盖在虚构公司的项目上——未读取工作区、注册表或键。同样是俄文;项目页面。

该发行版是完整的引擎：项目和仓库库存、发现、历史、指标、本地仪表盘、MCP、凭证工具以及可选的提供者集成。每个用户都提供自己的项目路径、账户和密钥。私有运营数据和 Git 历史被排除在源发行版之外。发布及其注释在发布页面;每个版本的变更内容都在变更日志中。

新队友快速起步

完整引擎支持 macOS 和 Linux，Python 3.11+ 和 SQLite 3.37+，并支持可加载扩展。在 Windows（Python 3.

Project Observatory

Your projects. Back in view. A local dashboard for the projects your agents work on. See what changed, what needs attention and where known API keys left a copy. In English or Russian. Project Observatory is an open-source tool from PassionCode.ai (product page), beside Fabric Switchboard. It is Fabric-compatible — its server speaks fabric-service/0.1 and its MCP fabric-interop/0.1 — and works without Fabric.

The real dashboard, rendered by tools/demo_estate.py over a fictional company’s projects — no workspace, registry or key was read. Also in Russian; the projects page.

The distribution is the complete engine: project and repository inventory, findings, history, metrics, a local dashboard, MCP, credential tools and optional provider integrations. Every user supplies their own project paths, accounts and keys. Private operational data and Git history are excluded from the source distribution. Releases and their notes are on the releases page; what changed in each is in the changelog.

Quick start for a new teammate

The complete engine supports macOS and Linux, Python 3.11+ and SQLite 3.37+ with loadable-extension support. On Windows (Python 3.13+) the engine, the CLI, the tick, the server and maintenance run (the background jobs as per-user Task Scheduler tasks, systemd user units on Linux) and are tested on every change, with the engine suites that do not pass on Windows yet named one by one (what is verified); a desktop app ships for Windows and Linux (below, design). The python3 that ships with macOS is 3.9, and pip run from it fails with a misleading ResolutionImpossible rather than naming the interpreter, so choose an extension-enabled interpreter such as Homebrew Python explicitly; the onboarding guide checks this. Every machine runs a tagged release, the same wheel byte for byte (staying in step).

Install

The latest release’s wheel, checked against its SHA256SUMS, into a virtual environment of its own:

V=$(curl -s https://api.github.com/repos/passioncode-ai/project-observatory-dashboard/releases/latest \
    | python3 -c 'import json,sys;print(json.load(sys.stdin)["tag_name"][1:])')
mkdir -p ~/observatory-release && cd ~/observatory-release