Hail Hydra

🐉 H Y D R A

  Multi-Headed Speculative Execution for AI Coding CLIs
  Claude Code  ·  Gemini CLI  ·  Codex CLI  ·  GitHub Copilot CLI

  "Cut off one head, two more shall take its place."
  Except here — every head is doing your work faster and cheaper.

  10 agents  ·  10 commands  ·  6 hooks  ·  4 host CLIs  ·  Codebase map  ·  Real token tracking  ·  Persistent memory

Hail Hydra is a free multi-agent toolkit for Claude Code, Gemini CLI, Codex CLI and GitHub Copilot CLI. It gives your coding CLI ten specialist agents, called heads. Each head is pinned to the cheapest model that can do its job well. Your main model plans, decides and verifies. The heads explore, run, write and check.

When you invoke it, Hydra typically cuts per-task cost by about 50%, and delegated work runs 2 to 3× faster. Your main model verifies the results, so quality holds.

- Ten heads, two tiers. Seven cheap-tier heads (Haiku, Flash-Lite, GPT-6 Luna) for search, tests, docs, git and scans. Three mid-tier heads (Sonnet, Flash, GPT-6 Sol) for code, review and deep checks.
- Sentinel verification. After substantial edits, a fast scan checks imports, signatures and dependencies. It escalates to a deep check only when something looks wrong, and it catches about 72% of integration bugs before runtime.
- Codebase map. A persistent dependency graph gives instant blast-radius lookups. Sentinel scans become 3 to 5× cheaper.
- Real receipts. /hydra:stats reads your CLI's own session logs and reports what you actually spent and saved. No AI estimation.
- Memory that compounds. Each head remembers your project's conventions, fragile spots and past false alarms across sessions.

One canonical source generates a…