SuperAPI AI Directory

super-log

Read-only access to a local dev bench that interleaves every log stream on one timeline: tail and search live streams, search the on-disk journal, wait for an event, and read per-topic guides for

Open website

About this tool

Purpose: converge log streams scattered across multiple places (devices, servers, containers, browsers, chains, applications) onto the same screen, interleaved by hub sequence, solving the problems of inconsistent timestamps across multiple terminals and issues that only appear at the interleaving points. Free and self-hosted.

Main capabilities: provides a unified small wire protocol (one JSON event per line, sent in batches via ordinary HTTP POST), the hub fans out to any number of readers and supports replay on connection; colors by source and level, can filter by stream, minimum level, or substring, supports pause, copy, and export to JSON/CSV/plain text. Provides more than a dozen dependency-free SDKs (C++, C, Rust, Python, Go, Java, Swift, Fortran, OCaml, Haskell, Lean 4, Perl, Lua, C#, Zig, COBOL, Ruby, shell, JS), some of which can hook into languages' existing logging frameworks. Supports cross-layer tracing (withTrace generates a correlation id), by default captures uncaught exceptions and unhandled rejections, host-side tailers capture adb logcat, iOS simulator, macOS unified logging, journald, Docker, and log files, and can pull entire batches of server logs over ssh. Also includes a log proxy, EVM address monitoring, gas balance guardian, DNS/TLS/port discrepancy monitoring, build eventization, large file download progress and stall monitoring, macOS power and temperature, crash reports and kernel panics, file change diffs, disk journal history retrieval and replay. GET /recent is for scripts and agents to read, and a built-in MCP server provides read-only access for coding agents (npx -y @super-log/mcp).

How to use: install with Homebrew on macOS, with a PPA on Ubuntu, directly with a .deb on Debian/Raspberry Pi OS, plus Fedora/RHEL, npm, vcpkg, and source build methods; the hub runs on :7333 by default; the superlog CLI is used to read, manage tailers, and push streams.

Limitations: analytics is a separate concern and can be left to super-log.com or your own storage; high-speed streams are rate-limited with an explicit notice rather than silently dropped; MCP is read-only by design (only the agent.* status topics have a narrow exception); remote configuration does not install software, does not open ports, and servers do not need to connect back to the hub; sensitive credentialsare always redacted.