Versions:
telltale, published by San Lee and currently at version 0.2.0 (the sole version listed), is a monitoring and comparison tool for coding-agent command-line interfaces, described by its author as an honest gauge for coding agents. It falls into the developer tooling category, specifically observability and telemetry for AI-assisted coding workflows. Its purpose is to let several coding-agent CLIs answer one brief side by side while standing on a statusline and a cross-vendor heads-up display in which every displayed number is traceable to measured tool output; a value that cannot be sourced is shown as absent rather than as a plausible zero, and any derived value is explicitly marked as derived. The entire product ships as a single binary. The central use case is `telltale council`, the dispatch room, which the publisher identifies as the product itself. Supporting it are two gauges: `telltale statusline`, an in-prompt statusline for Claude Code and Antigravity CLI, and `telltale hud`, a watch-mode terminal user interface spanning Claude Code, Codex CLI, Gemini CLI, Antigravity CLI, Cursor, and Grok CLI. The same binary also provides `telltale snapshot`, which exposes the fleet state as a single JSON document suitable for programmatic consumption, and `telltale doctor`, a launch-time preflight check over the vendor binaries, together with the relays that collect token counts and fleet events. Typical use cases include comparing how multiple coding agents handle the same brief, monitoring token usage and agent activity across vendors in real time, integrating agent fleet state into other programs, and verifying that vendor binaries are correctly installed before launch. Windows is the continuously verified target platform: every commit runs the test suite, the build, and binary-level smoke tests on windows-latest.
Tags: