Turn your coding agent into a content farm.

Open-source runtime for agent-driven content production. You chat with Hermes, Claude Code, Codex, or another local coding agent; the agent drives Ralphy. Forkable, observable, reproducible.

What it is

Ralphy is a tool for local agents, not a CLI you operate by hand. You stay in chat — Hermes, Claude Code, Codex, or another coding agent — and describe what you want; the agent runs Ralphy for you. The CLI is the runtime that gives the agent reproducible model calls, account workspaces, shared brand assets, content units, quality gates, renders, logs, and memory.

Under the hood, two API keys (OPENROUTER_API_KEY + ELEVENLABS_API_KEY) wire up image / video / vision / LLM (OpenRouter), voice + music (ElevenLabs), HTML+GSAP composition (HyperFrames), and a local async-job queue (bun + SQLite). Direct ralphy <verb> commands stay available for setup, debugging, and power users — but driving them yourself is not the primary workflow.

Demo

See what Ralphy makes → — real rendered outputs from real projects.

Cost: ~$8–12 per 30s video. Speed: ~8 min cold-start, ~25 min for a 10-batch. Engine: HyperFrames (HTML + GSAP, deterministic Puppeteer + FFmpeg render).

Install

All four ship the same binary.

Setup is a one-time step you run once so your agent can take over from there. These two commands are agent enablement + diagnostics, not the everyday workflow:

ralphy setup          # interactive wizard — paste the two API keys + install agent skill
ralphy doctor         # verify env is green (run this when the agent reports a problem)

Expected output:

✦ ralphy v0.3.0
▸ Dependencies          ✓ bun  ✓ ffmpeg
▸ API keys              ✓ OPENROUTER_API_KEY  ✓ ELEVENLABS_API_KEY
  ✓ ready

macOS Gatekeeper warning? You used the direct-download path. Brew / npm / install.sh bypass Gatekeeper automatically. If you hit it: xattr -d com.apple.quarantine /path/to/ralphy once and you're done.

Verify your install: every Release includes a SHA256SUMS file. shasum -a 256 -c SHA256SUMS (macOS / Linux) or Get-FileHash (Windows) confirms the binary matches.

60-second tour

In practice you say "make a spring espresso ad" in chat and the agent runs these verbs for you. Here's the surface it drives, so you can see what's happening under the hood:

# 1. Create a project
ralphy new "Spring espresso ad" --id espresso-001

# 2. Find a template by free-text utterance
ralphy template suggest "talking head rant about deadlines" -p

✦ Query: "talking head rant about deadlines"
  1. ✓ talking-head  ███████████████░  0.95  strong
  2. ✓ story-time    ███████████░░░░░  0.70  strong

# 3. Scaffold from the chosen template (sourced from the hosted library)
ralphy template use talking-head --id espresso-001

# 4. Cost-preview before spending a cent
ralphy generate image --project espresso-001 --slot scene-01-bg \
  --prompt "studio packshot, white seamless, 50mm, photoreal" --dry-run

# 5. Render the project to mp4
ralphy render espresso-001

That's it. Full CLI surface in docs/cli-surface.generated.md.

Why Ralphy

What you actually get vs other ways to do this. The operator is your agent; you stay in chat.

The hard rule that makes the rest work: ralphy <verb> is the only entry-point. No ad-hoc ffmpeg shell-outs, no direct provider fetches, no orphan scripts. Every model call lands in generations.jsonl, every cost in the rollup, every failure in the postmortem.

Architecture

graph LR
    A[Agent: Claude Code / Cursor / Codex] -->|playbooks| B[ralphy CLI]
    B --> C[Provider router]
    C --> D[OpenRouter<br/>Kling / Seedance / Veo / Sora / Nano-Banana]
    C --> E[ElevenLabs<br/>TTS + Music]
    B --> F[HyperFrames composer<br/>HTML + GSAP]
    F --> G[mp4 via Puppeteer + FFmpeg]
    B --> H[Project memory<br/>genlogs · postmortems · cost rollup]
    B --> I[Hosted template library<br/>+ guidelines in git]

5 agent roles (researcher / scenarist / art-director / editor / producer) routed via AGENTS.md. The router decides which playbook the agent reads before acting.

Documentation & community

Contributing

git clone https://github.com/alecs5am/ralphy.git
cd ralphy && bun install

bun test                       # unit + integration (1,000+ tests)
bun run lint                   # typecheck + project lints (errors / help-examples / skills / agents-md / cli-surface)
bun run cli:surface:build      # regenerate docs/cli-surface.generated.md
bun run build:bin              # build cross-platform binaries

A pre-commit hook runs the test suite. CI runs the same on push/PR.

PRs welcome — especially:

• New model entries in MODELS.md with real cost numbers + known pitfalls.
• Bug fixes in cli/lib/providers/.
• New guidelines under guidelines/<slug>/ (image-prompt rules — tag-able from chat as @guideline:<slug>).

For non-trivial changes, open an issue first or start a discussion.

License

Apache 2.0. Use, fork, ship to prod — patent grant included.