Back to directory

Albatross

A terminal-first coding agent that lets you bring your own API key or point at a local model, with transparent multi-model routing, per-turn cost display, session branching, and a separate critic agent that grades work against a rubric.

Albatross

Albatross is a small coding harness that runs in the terminal and works the same way whether it is talking to a hosted API or a local model server such as Ollama, LM Studio, MLX, or llama.cpp: the same tools, commands, and session log apply either way, and a provider can be switched mid-session.

It shows per-turn and per-session cost on the status line when pricing is known, scores and routes among configured coder models, and can break a task into a graded plan that a configured planner model executes tier by tier. An iteration loop pairs a generator with a separate critic agent that scores each pass against a rubric until it clears the bar, rather than letting the generator grade its own work.

Other features include real turn-level undo, forking a session into two paths to compare fixes, and MCP server support alongside its built-in read, edit, grep, shell, and test tools.

More in Coding agents

See all 793
  • The City

    github.com

    A native macOS app that shows Claude Code runs as a city, with each project as a building, each team as a floor of robots, and every animation driven by a real event from the claude CLI's stream.

  • rig

    tryrig.ai

    A small, single-binary agent runtime that points at any OpenAI-compatible model, local or hosted, reads and edits a repository's code, runs its tests and remembers what it learned, with a loop implemented in a few hundred lines.

  • Jauvex

    jauvex.reindent.com

    An Electron desktop app that puts Claude Code, Codex and Grok coding-agent sessions side by side and lets you steer them by voice, with a small model handling fast routing decisions.