Gaia is a living, open registry of every AI agent skill — structured as a leveling skill graph. Live site: https://gaiaskilltree.com/

Research backbone for gaia-skill-tree

Multi-harness token-usage cost reporter for AI agent sessions (pi, Claude Code, Codex, opencode). Reads JSONL logs, prices from LiteLLM catalog, auto-refresh built-in. One file, zero deps.

AI agent command consolidator for Claude Code, Cursor, Windsurf. Merge overlapping /commands into one unified SKILL.md — reduce slash command clutter, combine Cursor rules, dedupe Windsurf cascades. One-line install.

Dependency-free browser runtime for compiled Milim character releases.

Strip your agent's context bloat — run clean. Skill Heaven: per-harness clean-room doors (claude-heaven, pi-heaven, …) + the Hell/Heaven benchmark research driver.

Agent-native discovery, trust, installation, and progression interface for the Gaia Skill Tree

GitHub Actions cost analyzer & flaky-test detector for pull requests. Measures wasted CI compute, classifies commits as feature vs ci-fix, generates pre-push checks that would have prevented the retry-push churn. One Python file, gh CLI, no config.

Measure + compact an oversized agent-context file (CLAUDE.md, .cursorrules, AGENTS.md, system prompt) under the harness char limit without losing a rule. /context-diet — output of Gaia Research Context Diet Lab 001.