WaveSpeed

wavespeed-pp-cli

Run WaveSpeed AI models with schema-aware inputs, prompt shortcuts, price estimates, wait/download helpers, aliases, and media uploads.

Printed by @cathrynlavery (Cathryn Lavery)

Install

npx — recommended

npx -y @mvanhorn/printing-press-library install wavespeed

Installs the CLI binary and the agent skill (Claude Code, Codex, Cursor, Gemini CLI, Copilot, and more). Add --cli-only or --skill-only for one half. Requires Node.

Hermes

hermes skills install mvanhorn/printing-press-library/cli-skills/pp-wavespeed --force

From the Hermes CLI. Inside a chat session use /skills install with the same path.

OpenClaw

Install the pp-wavespeed skill from https://github.com/mvanhorn/printing-press-library/tree/main/cli-skills/pp-wavespeed. The skill defines how its required CLI can be installed.

Paste this to your OpenClaw agent — the skill describes how to install its CLI.

Claude Desktop

This CLI ships an MCPB bundle. Download the latest release .mcpb and double-click it — Claude Desktop walks you through the install.

View source on GitHub →

Run any WaveSpeed model from the terminal, with price checks, safe uploads, and recovery-safe downloads.

WaveSpeed hosts image, video, audio, and 3D models behind one API. This CLI adds a dynamic model runner for slash-delimited model IDs, free price estimates, local file uploads, a generation library with cost reports, and a D2C content layer (plan, pack, batch, variants, compose, aspects, restyle, brand, qa). Paid submissions are never replayed, and every post-submit failure keeps the prediction ID and a recovery command.

Learn more at WaveSpeed.

Created by @cathrynlavery (Cathryn Lavery).

Authentication

WaveSpeed uses an API key sent as Authorization: Bearer <key>. Create one at https://wavespeed.ai/accesskey and export it as WAVESPEED_API_KEY. doctor verifies the key with a free /balance read.

Quick Start

# Verify the API key with a free balance read
wavespeed-pp-cli doctor

# Find an image-edit model and its price
wavespeed-pp-cli models --capability image-edit --agent

# Quote the cost before spending
wavespeed-pp-cli price --model-id google/nano-banana-2/edit --set aspect_ratio=9:16 --set resolution=2k --agent

# Generate, wait, and download (local files work too: --images @ref.png)
wavespeed-pp-cli run --model-id google/nano-banana-2/edit --prompt "a red mug on oak" --images https://wavespeed.ai/ref.png --set aspect_ratio=9:16 --wait --download 'out/mug_{index}.{ext}' --record --agent

Unique Features

These capabilities aren't available in any other tool for this API.

Core

run — Submit WaveSpeed model runs with prompt shorthand, typed --set inputs, local @file media uploads, price estimates, waiting, recovery-safe downloads, and library recording.

Use it for any single generation; the prediction ID and recovery command survive every post-submit failure.
```
wavespeed-pp-cli run --model-id google/nano-banana-2/edit --prompt "a red mug on oak" --images @ref.png --set aspect_ratio=9:16 --wait --download 'out/mug_{index}.{ext}' --agent
```
schema — Fetch the live WaveSpeed model catalog and print the request schema for a model or project alias.

Check accepted inputs and enums before a paid run.
```
wavespeed-pp-cli schema google/nano-banana-2/edit --agent
```
price — Estimate WaveSpeed model pricing with the same input syntax used by run, without submitting a prediction.

Free; quote the cost before spending.
```
wavespeed-pp-cli price --model-id google/nano-banana-2/edit --set aspect_ratio=9:16 --set resolution=2k --agent
```
aliases — Read wavespeed.json aliases, default model, and output directory settings for repeatable local workflows.

See which short names map to which models.
```
wavespeed-pp-cli aliases --agent
```
init — Write a starter wavespeed.json with aliases, default model, and output directory.

Start a new image project.
```
wavespeed-pp-cli init
```

Media

upload — Upload local image, video, or audio files to WaveSpeed media storage for use as model input URLs, retrying stalled uploads with a size-scaled deadline.

Turn a local file into a URL a model accepts.
```
wavespeed-pp-cli upload ./ref.png --agent
```
download — Download generated output URLs with directory, exact-path, or templated-path destinations, never sending API credentials to CDN hosts.

Re-fetch outputs from a finished prediction.
```
wavespeed-pp-cli download "$OUTPUT_URL" --output 'out/{index}.{ext}'
```
last — Print or open the most recent downloaded output.

Grab the path of the last generated file.
```
wavespeed-pp-cli last
```

Plan

plan brief-to-shotlist — Turn a free-text brief into a structured shotlist across platforms and aspect ratios with a hybrid deterministic-parser/LLM planner.

Draft the shot list for a campaign.
```
wavespeed-pp-cli plan brief-to-shotlist --prompt "Helm Black launch" --platforms instagram,tiktok --agent
```
plan model-pick — Recommend a model for an intent from the live catalog with rationale.

Choose a model for a job.
```
wavespeed-pp-cli plan model-pick "short product video" --agent
```
plan cost-estimate — Price a shotlist against live /model/pricing and the account balance, with cached-pricing fallback and per-shot breakdown.

Check a shotlist fits the budget.
```
wavespeed-pp-cli plan cost-estimate shotlist.json --agent
```
qa preflight — Pass/warn/fail validation of a shotlist: balance vs cost, model availability, prompt safety, platform request-shape, and brand coverage.

Gate a pack before producing it.
```
wavespeed-pp-cli qa preflight shotlist.json --agent
```

Produce

pack — Produce a multi-platform creative pack from one concept at stable packs/// paths with per-platform manifests, concurrency, cost ceiling, and image-dimension validation; a rerun archives the superseded manifest.

Produce platform-ready assets.
```
wavespeed-pp-cli pack --concept "Helm Black hero" --platforms instagram,tiktok --max-cost 5.00 --agent
```
batch — Submit many prompts from CSV or JSON with a spend ceiling and fail-fast/fail-tolerant semantics; records completed generations before any abort.

Run a prompt list under a budget.
```
wavespeed-pp-cli batch --from prompts.csv --max-cost 5.00 --agent
```
variants — Sweep seed, style, or model off a base shot to produce comparable outputs with side-by-side metadata.

Explore seeds or models for one shot.
```
wavespeed-pp-cli variants --base shotlist.json --vary seed --count 4 --agent
```
compose — Run an explicit multi-step pipeline (text->image->upscale->video), feeding each step's output to the next, with rollback of later steps on failure.

Turn a prompt into an image and then a clip.
```
wavespeed-pp-cli compose --steps "text->image,image->video" --prompt "..." --models m1,m2 --agent
```

Refine

aspects — Re-frame one image into standard platform aspect ratios, using outpaint when supported and an anchored re-render otherwise.

Make 9:16 and 1:1 versions of a hero image.
```
wavespeed-pp-cli aspects hero.png --platforms instagram,tiktok --agent
```
restyle — Apply a brand profile or explicit style to an existing asset via img2img with a style prompt.

Bring an asset onto brand.
```
wavespeed-pp-cli restyle hero.png --brand helm --agent
```

Library

library — List, search (FTS5), show, tag, export, and cost-report the local generation library by brand, model, platform, and tag.

Find a past generation or total spend.
```
wavespeed-pp-cli library cost-report --group-by model --agent
```
brand — Create, inspect, apply, and edit brand profiles that auto-merge into pack, compose, variants, restyle, and run.

Set up a brand for consistent output.
```
wavespeed-pp-cli brand init helm --palette '#111,#eee' --voice calm
```

Recipes

Image edit with a reference, 9:16 at 2k

wavespeed-pp-cli run --model-id google/nano-banana-2/edit --prompt "$PROMPT" --images https://wavespeed.ai/anchor.png --set aspect_ratio=9:16 --set resolution=2k --wait --download 'raw/shot_{index}.{ext}' --record --agent

Pass a URL or a local @file; local files upload automatically. The download template names each output.

First-and-last-frame video

wavespeed-pp-cli run --model-id bytedance/seedance-v1.5-pro/image-to-video --prompt "$MOTION" --image https://wavespeed.ai/a.jpg --last-image https://wavespeed.ai/b.jpg --set duration=5 --wait --wait-timeout 10m --download 'clip_{index}.{ext}' --record --agent

--wait-timeout bounds polling; on timeout the prediction ID and recovery command are printed.

Spend for a session

wavespeed-pp-cli billings --page-size 20 --agent

Billing rows carry the charged price per prediction.

Plan a campaign

wavespeed-pp-cli plan brief-to-shotlist --prompt "launch" --platforms instagram,tiktok --agent

Save the shotlist, then run plan cost-estimate and qa preflight on it before pack.

Usage

Run wavespeed-pp-cli --help for the full command reference and flag list.

Paths & environment variables

This CLI separates local files into four path kinds:

Kind	Contents
`config`	User-editable settings such as `config.toml` and saved profiles
`data`	Durable local data: `credentials.toml`, `data.db`, cookies, browser-session proof files, and other auth sidecars
`state`	Runtime state such as persisted queries, jobs, and `teach.log`
`cache`	Regenerable HTTP/cache files

Each kind resolves independently. The ladder is:

Per-kind env var: WAVESPEED_CONFIG_DIR, WAVESPEED_DATA_DIR, WAVESPEED_STATE_DIR, or WAVESPEED_CACHE_DIR
--home <dir> for this invocation
WAVESPEED_HOME for a flat relocated root
XDG env vars: XDG_CONFIG_HOME, XDG_DATA_HOME, XDG_STATE_HOME, XDG_CACHE_HOME
Platform defaults matching existing installs

For containers and agent sandboxes, prefer a single relocated root:

export WAVESPEED_HOME=/srv/wavespeed
wavespeed-pp-cli doctor

Under WAVESPEED_HOME=/srv/wavespeed, the four dirs resolve to /srv/wavespeed/config, /srv/wavespeed/data, /srv/wavespeed/state, and /srv/wavespeed/cache.

MCP servers do not receive CLI flags from the host. Put relocation in the host env block:

{
  "mcpServers": {
    "wavespeed": {
      "command": "wavespeed-pp-mcp",
      "env": {
        "WAVESPEED_HOME": "/srv/wavespeed"
      }
    }
  }
}

Precedence matters in fleets: an ambient per-kind variable such as WAVESPEED_DATA_DIR overrides an explicit --home for that kind. Use WAVESPEED_HOME or the per-kind variables for durable fleet relocation; treat --home as the weaker per-invocation lever.

Relocation is one-way. Unsetting WAVESPEED_HOME does not move files back to platform defaults, and doctor cannot find credentials left under a former root. Move the files manually before unsetting relocation variables.

Existing installs keep working because the platform-default rung matches the legacy layout. On the first auth write, stored secrets leave config.toml and are consolidated into credentials.toml under the data directory. Run wavespeed-pp-cli doctor --fail-on warn to check path and credential-location warnings in automation.

Commands

account_balance

Manage account balance

wavespeed-pp-cli account-balance - Retrieve the authenticated account balance.

billings

Billing and usage records

wavespeed-pp-cli billings - Search billing records for the authenticated account.

media_uploads

Manage media uploads

wavespeed-pp-cli media-uploads - Upload one existing local media file (image, video or audio) to WaveSpeed storage and return its URL for model inputs.

model_pricing

Manage model pricing

wavespeed-pp-cli model-pricing - Estimate the unit price for a model run using the same inputs that will be submitted to the model endpoint.

models

Model catalog and model metadata

wavespeed-pp-cli models - List available WaveSpeed models and their API schemas.

prediction_deletions

Manage prediction deletions

wavespeed-pp-cli prediction-deletions - Delete one or more predictions from history.

prediction_results

Manage prediction results

wavespeed-pp-cli prediction-results <task_id> - Retrieve the latest status and result payload for a prediction task.

predictions

Prediction submission history and result retrieval

wavespeed-pp-cli predictions - Query recent prediction history. The API history window is limited; sync accumulates across runs.

usage_stats

Manage usage stats

wavespeed-pp-cli usage-stats - Retrieve usage statistics for the authenticated account.

Self-learning loop

This CLI caches per-question discovery so repeat queries skip the walk and structurally similar queries get answered via entity substitution. The loop also self-captures: every invocation is journaled locally, and failed-flag corrections plus fresh teaches surface as candidates on the next recall for confirm/reject judgment. Agents call recall before discovery and fire teach & after answering. See the ## Automatic learning section in SKILL.md for the full protocol.

wavespeed-pp-cli recall <query> - Look up cached resources for a query before running discovery
wavespeed-pp-cli teach - Record a query -> resource mapping (silent on success, safe to background with &)
wavespeed-pp-cli learnings list - Inspect taught rows
wavespeed-pp-cli learnings forget <query> - Undo a teach
wavespeed-pp-cli learnings candidates - List auto-captured candidates awaiting confirm/reject
wavespeed-pp-cli learnings stats - Local loop metrics: recall hit rate, teach-to-reuse, playbook resolution, candidate counts
wavespeed-pp-cli teach-pattern - Install a query/resource template up front
wavespeed-pp-cli teach-lookup - Add an entity mapping (e.g. country code, team alias) for pattern substitution

Pass --no-learn or set WAVESPEED_NO_LEARN=true to disable the loop for deterministic flows.

The local store's schema version stamp is one-way: once this version of wavespeed-pp-cli opens the database, older binaries refuse it with a version error — upgrade the binary rather than downgrading.

Output Formats

# Human-readable table (default in terminal, JSON when piped)
wavespeed-pp-cli billings

# JSON for scripting and agents
wavespeed-pp-cli billings --json
# Filter to specific fields
wavespeed-pp-cli billings --json --select code,data,message

# Dry run — show the request without sending
wavespeed-pp-cli billings --dry-run

# Agent mode — JSON + compact + no prompts in one flag
wavespeed-pp-cli billings --agent

Agent Usage

This CLI is designed for AI agent consumption:

Non-interactive - never prompts, every input is a flag
Pipeable - --json output to stdout, errors to stderr
Filterable - --select <field>[,<field>...] returns only fields you need
Previewable - --dry-run shows the request without sending
Explicit retries - add --idempotent to create retries when a no-op success is acceptable
Explicit confirmation - --agent does not imply --yes; pass --yes separately only after the target, arguments, and side effects are clear
Piped input - write commands can accept structured input when their help lists --stdin
Offline-friendly - sync/search commands can use the local SQLite store when available
Agent-safe by default - no colors or formatting unless --human-friendly is set

Exit codes: 0 success, 2 usage error, 3 not found, 4 auth error, 5 API error, 7 rate limited, 10 config error.

Health Check

wavespeed-pp-cli doctor

Verifies configuration, credentials, and connectivity to the API.

Configuration

Run wavespeed-pp-cli doctor to see the resolved config, data, state, and cache directories. The platform-default config path is ~/.config/wavespeed-pp-cli/config.toml; --home, WAVESPEED_HOME, and per-kind env vars can relocate it.

Static request headers can be configured under headers; per-command header overrides take precedence.

Environment variables:

Name	Kind	Required	Description
`WAVESPEED_API_KEY`	per_call	Yes	Set to your API credential.

agentcookie (optional)

If you use agentcookie to sync secrets across machines, this CLI auto-adopts agentcookie-managed credentials with no extra setup. When the daemon writes to this CLI's config, wavespeed-pp-cli doctor reports agentcookie: detected and auth-status labels the source as agentcookie. Skip this section if you don't use agentcookie - the CLI works the same as any other.

Troubleshooting

Authentication errors (exit code 4)

Run wavespeed-pp-cli doctor to check credentials
Verify the environment variable is set: echo $WAVESPEED_API_KEY Not found errors (exit code 3)
Check the resource ID is correct
Run the list command to see available items

API-specific

A run timed out or the download failed after submission — The prediction may still be running and is billed once. Check it with wavespeed-pp-cli prediction-results "$PREDICTION_ID"; the ID is printed with every post-submit failure. A failed download after a completed run is a warning (exit 6, partial_failure), not a failed generation.
Upload of a large local file stalls — Uploads retry network errors, 429, and 5xx up to three times with a per-attempt deadline of --timeout plus size / 128 KiB/s. Uploads are free, so retries cannot add cost.
price quotes more than billings charged — price returns list price; billings shows the charged price after account discounts (order.price vs order.origin_price).