An agentic development harness for Claude Code. One command runs the entire feature pipeline, it collaborates with you on the design, then builds autonomously: plan, execute, double-review, and verify.
Claude Code
> /plugin marketplace add jdanigo/hydraia> /plugin install hydraia@hydraia
Codex
$ npx skills add jdanigo/hydraia -a codex
Cursor
$ npx skills add jdanigo/hydraia -a cursor
Gemini CLI
$ npx skills add jdanigo/hydraia -a gemini-cli
GitHub Copilot
$ copilot plugin marketplace add jdanigo/hydraia$ copilot plugin install hydraia@hydraia
OpenCode
$ npx skills add jdanigo/hydraia -a opencode
Windsurf
$ npx skills add jdanigo/hydraia -a windsurf
🇬🇧 English · 🇪🇸 Español
An agentic development harness for Claude Code. One command runs the entire feature pipeline — it collaborates with you on the design, then builds autonomously: plan, execute, double-review, and verify. No per-step babysitting, no choosing which model or skill to use.
/hydraia:feature add rate limiting to the public REST API
You stay on Opus; Hydraia decides the rest — how much ceremony the change needs
(three auto-selected levels), when to drop to Sonnet for execution, which reviewers
to run, which security gates to enforce — and every run ends DONE,
DONE_WITH_FOLLOWUPS, or BLOCKED with evidence, never in an endless fix loop.
Raw Claude Code is powerful but hands-on: you remember to plan, to threat-model, to review, to test — every step manual, every step skippable under pressure. Hydraia applies the discipline you'd use on your best day, on every run. One command runs a fixed pipeline — think → design + threat model → plan → execute → double review → verify — with every model and reviewer decision already made, and a security gate that is always on and never configurable.
| Phase | What happens | Model |
|---|---|---|
| Design | Brainstorm → spec + threat model (scaled by level) | Opus |
| Plan | Plan-as-contract (Files + Verify per task), then frozen | Opus |
| Execute | Fresh sub-agent per task, scope-gated | Sonnet (Haiku for mechanical) |
| Review | Judge + diff-scoped reviewers + security gate → verified triage | Opus judges, Sonnet reviewers |
| Verify | Baseline-aware tests, spec check, secrets/deps scan — bounded | Opus |
Prerequisites (install before Hydraia): Claude Code, git, Node.js ≥18, and
Python 3.8+. codegraph and markitdown are installed for you by
/hydraia:doctor. Platform: macOS/Linux (Windows via WSL).
Add the marketplace and install the plugin — inside the Claude Code CLI or your terminal:
claude plugin marketplace add jdanigo/hydraia
claude plugin install hydraia
Then run /hydraia:doctor once to validate and install external deps. That's it —
every skill and agent ships inside the plugin.
| Command | What it does |
|---|---|
/hydraia:feature <desc> | Full pipeline: context → design → plan → build → double review + security → verify |
/hydraia:plan <desc> | Design + threat model + detailed plan, then stop (nothing executed) |
/hydraia:agile <idea> | Decompose an epic into stories/tasks and deliver it autonomously, stage by stage |
/hydraia:review [focus] | Double review + security gate on the current branch |
/hydraia:resume | Continue an interrupted run from the last incomplete phase |
Build & plan
| Command | What it does |
|---|---|
/hydraia:feature <desc> | Full pipeline end to end: context → think → design + threat model → plan → execute → double review + security gate → verify |
/hydraia:plan <desc> | Context + design + threat model + detailed plan (with self-review), then stops. Nothing is executed |
/hydraia:agile <idea> | Multi-stage autonomous delivery: decomposes an epic into stories → tasks and executes the tree by dependency, gated by autonomy tier |
/hydraia:story <story> | Product-owner analysis of a user story (INVEST, acceptance criteria) → spec → QA cases + traceability matrix → frozen plan, then stops |
/hydraia:architect <idea> | Greenfield: guided elicitation → architecture options → confirmed stack → API contract → ADRs → full build pipeline |
Diagnose & specialize
| Command | What it does |
|---|---|
/hydraia:perf <symptom> | Measurement-first performance run: baseline → profile-driven diagnosis → numeric target → implement → re-measure |
/hydraia:db <symptom> | DB bottleneck run: engine detection, read-only evidence (EXPLAIN, stats, locks), expand-contract migrations |
/hydraia:e2e [focus] | Generate + run a Playwright critical-flow E2E suite (auto-detected framework) |
/hydraia:devops <request> | Author CI/CD, Docker, or IaC — deploy and secrets steps flagged for human approval |
/hydraia:observability <request> | Instrument logs / metrics / traces / alerts — OTel-first, never logs secrets or PII |
Review, understand & docs
| Command | What it does |
|---|---|
/hydraia:review [focus] | Double code review + security gate on the current branch (code already exists) |
/hydraia:graph <query> | Query the code graph — call sites, blast radius — without running the pipeline |
/hydraia:explainme <focus> | Interactive HTML map of a codebase, subsystem, or PR (onboarding in 10 min) |
/hydraia:docs [focus] | Sync README, API docs, CHANGELOG, and ADR index with the code — reports drift |
Utilities
| Command | What it does |
|---|---|
/hydraia:resume [run] | Continue an interrupted pipeline from the last incomplete phase |
/hydraia:doctor | Validate, install, and update external deps (codegraph, markitdown), with consent |
/hydraia:dashboard [port] | Launch a local (127.0.0.1) web dashboard: plugin status, usage telemetry, editable run modes |
Ship a feature, end to end
/hydraia:feature add rate limiting to the public REST API — 100 req/min per key
Planned, built, reviewed twice, security-gated, and verified — one command.
Lock the approach before writing code
/hydraia:plan migrate the session cache from in-memory to Redis
You get a spec + threat model + frozen plan and nothing is executed. Run
/hydraia:feature when you're happy.
Review a branch you didn't build with Hydraia
/hydraia:review the auth module
Whole-branch + diff-scoped review with the mandatory security gate.
Understand blast radius before touching anything
/hydraia:graph what calls parseConfig and what breaks if I change its signature
Turn a codebase into an onboarding map
/hydraia:explainme the payments subsystem
Questions, ideas, showcases, and release news live in the Hydraia Discord — come say hi:
npx skills (skills/<name>/SKILL.md layout)codex/)Stars on GitHub
Helps you take a change from a rough idea or bug report through to a tested, reviewed result, with each step handled by a focused skill. Its skills cover pinning down domain terms, designing module interfaces, diagnosing hard bugs, and reviewing changes against your repo's coding standards.
@mattpocockA development methodology for coding agents that makes the agent brainstorm before building, follow plans, run parallel work and finish branches in a disciplined way. You get fewer skipped steps and more consistent results from your coding assistant.
@obraGives your coding agent a planning, testing and review routine to follow on every task. It plans before building, checks changes with tests, has a fresh reviewer look for regressions, and saves repeated wins as reusable skills.
@affaan-mDeveloper tools for working on the React codebase itself. Provides skills for extracting error messages, managing and debugging feature flags across release channels, running Flow type checks, and fixing lint and formatting issues before committing.
@facebookAn agentic development harness for Claude Code. One command runs the entire feature pipeline, it collaborates with you on the design, then builds autonomously: plan, execute, double-review, and verify. Install it with the command on its page. It includes hydraia, adr, agency-tier-ui. Its skills do not fire on their own yet. Free and open source.
Is this plugin yours?
Claim it with GitHubSubmit a pluginPromote it