dreamina-canvas-plugin

Dreamina Canvas

Plugin marketplaces

This plugin belongs to AIGC content creation.

Category · Marketplace · Purpose Full-stack development · Full Stack Plugins · Architecture and UI design, code understanding, quality checks, code review, workflow governance, and server operations AIGC content creation · Full AIGC Plugins · Image, video, audio, music, 3D, and multimodal content creation

Build Dreamina canvases from typed nodes and a timeline, keep the free save step separate from the quoted run, and download assets with byte counts and SHA-256 receipts.

English | 简体中文 · Install · Quick start · Skills · Architecture · Troubleshooting

Positioning

dreamina-canvas turns a brief into a structured Dreamina canvas: typed nodes, a timeline, capability discovery at run time, a free draft save, then a quoted run that only executes after you approve it. The plugin never hard-codes catalog values, never merges the save step with the run step, and never mints a replacement submitId to recover from a timeout.

It is a strict wrapper around the installed dreamina-canvas CLI and ships no MCP server: the eleven Skills drive the CLI through an argv-only adapter. Authentication and canvas state stay with the CLI; the plugin persists only non-secret operation identifiers.

Installation

Prerequisites

• Python 3.11 or newer on PATH.
• The dreamina-canvas CLI, installed and authenticated before first use. The setup and auth Skills can run the official installer and login, but only after explicit authorization, and the binary is never bundled.
• Visual-loop runtime dependency: jsonschema from requirements.txt. Install with python -m pip install -r requirements.txt.
• Development and test gates additionally use requirements-dev.txt.

From the plugin marketplace

codex plugin marketplace add partme-ai/full-aigc-plugins
codex plugin add dreamina-canvas@full-aigc-plugins

Restart the host and open a new task so the Skills load. Then verify the packaged Skills still match the upstream lock file:

python scripts/verify_dreamina_canvas_skills.py

Quick start

flowchart LR
    A[Brief] --> B[Discover live models and voices]
    B --> C[Compose nodes and timeline]
    C --> D[Save draft, no charge]
    D --> E[Quote the exact node set]
    E --> F{Approval}
    F -->|approved| G[Run once]
    F -->|rejected| C
    G --> H[Query by submitId]
    H --> I[Download with byte and SHA-256 receipts]

Create a Dreamina canvas for a 15-second product teaser. Draft the nodes and
do not run anything yet.

Expected observation: live capabilities are discovered, typed nodes are composed, a draft is saved, and the plugin confirms that nothing was charged. Ask for a quote next; the run happens once, after you approve the credit total.

What you can build

Workflow · Built-in path · Verifiable output Image nodes · dreamina-canvas-cli-text2image, dreamina-canvas-cli-image2image · Saved draft, quoted run, verified image assets Video nodes · dreamina-canvas-cli-text2video, dreamina-canvas-cli-ref2video · First/last-frame or reference-driven video assets Audio nodes · dreamina-canvas-cli-text2voice, dreamina-canvas-cli-text2audio · Narration or music assets with the chosen voice Canvas composition · dreamina-canvas-cli, /dreamina-canvas-compose, /dreamina-canvas-timeline · Node graph, timeline tracks, replaced clips and audio Operate and recover · dreamina-canvas-cli-auth, /dreamina-canvas-run, /dreamina-canvas-resume · Approval record, quote, ledger entry and resumed operation

Skills

Skill · What it does dreamina-canvas-cli · Shared canvas operations: model and voice discovery, canvas and asset management, node editing, timeline, quote-bound run, resume and verified download. dreamina-canvas-cli-setup · Installs, updates or verifies the CLI, including PATH troubleshooting. dreamina-canvas-cli-auth · Login, device authorization, account check, credential refresh and logout. dreamina-canvas-cli-text2image · Text-to-image mode: inputs, validation, draft and scene acceptance. dreamina-canvas-cli-image2image · Image-to-image and upscale mode. dreamina-canvas-cli-text2video · Text-to-video mode. dreamina-canvas-cli-ref2video · Reference-media video, including m2v and first/last-frame modes. dreamina-canvas-cli-text2voice · Text-to-speech mode. dreamina-canvas-cli-text2audio · Music generation mode. dreamina-canvas-harness · Routing and controller boundary for approvals, one-round visual evaluation, recovery and delivery evidence. dreamina-canvas-use · Routes an unspecified image, video, voice or music request to the right entry point.

Slash commands

Command · What it does /dreamina-canvas · Entry point; routes canvas creation, composition and generation. /dreamina-canvas-setup · Install, upgrade or verify the CLI, including PATH fixes. /dreamina-canvas-auth · Check, refresh or recover the login state. /dreamina-canvas-create · Create or switch a canvas and confirm the projectId. /dreamina-canvas-image · Text-to-image: draft, approved run, verified output. /dreamina-canvas-image2image · Image-to-image and upscale. /dreamina-canvas-video · Text-to-video: draft, approved run, verified output. /dreamina-canvas-ref2video · Reference-media or first/last-frame video. /dreamina-canvas-audio · Text-to-speech with a selected voice. /dreamina-canvas-music · Music generation by style and duration. /dreamina-canvas-assets · Upload assets, check status and place them on the canvas. /dreamina-canvas-compose · Compose text, element and generation nodes in dependency order. /dreamina-canvas-timeline · Arrange tracks, replace clips and audio. /dreamina-canvas-run · Quote saved nodes, obtain credit approval and submit once. /dreamina-canvas-resume · Query, wait for or resume a submitted operation.

Configuration

There is no plugin configuration file and no credential in this repository. Behaviour is driven by CLI arguments plus your explicit approval.

Setting · Where it lives · Notes CLI binary · dreamina-canvas on PATH · Overridable through the adapter for controlled deployments Authentication · The CLI's own store · The plugin never reads or writes it Operation ledger · /operations//.json · Mode 0600; non-secret fields only Credit ceiling · Supplied with the approval · A ceiling below the latest authoritative total is refused

Architecture and security

dreamina-canvas-plugin owns the adapter, the approval guard, the operation ledger and the artifact guard. The CLI owns login and remote state; Dreamina owns the canvas service. The authors receive no asset.

• Save and run are separate operations; no quote means no approval is possible.
• An approval is bound to the quoted node set and its project, and a consumed approval is rejected as a replay.
• Timeouts resume through the existing submitId; a missing ledger record escalates to a human instead of retrying.
• Only non-secret identifiers are persisted, and credential-like field names are rejected before anything is written.

Evidence · Status Offline, unit and scenario gates · Current local regression record in docs/verification/round-safety-2026-10-08.md Visual single-round controller · Target upload and one-credit paid canary recorded for the 2026-09-22 controller Live, paid and Windows reruns for 0.5.0 · Not claimed by this README; see docs/verification/ for the dated records

Troubleshooting

Symptom · Check first · Resolution Skills are missing · Plugin status · Restart the host and open a new task The CLI is not found · dreamina-canvas on PATH · Run /dreamina-canvas-setup and approve the install, or install it manually A run pauses for approval · The pending quote · Review the credit total, then approve or reject A run times out · The ledger entry · Resume by the existing submitId; do not re-quote blindly A download fails verification · The receipt · Re-download; a mismatched byte count or digest is a hard failure Login is required · The CLI session · Authenticate with the CLI, then retry

Upgrade

codex plugin marketplace upgrade full-aigc-plugins
codex plugin add dreamina-canvas@full-aigc-plugins

Re-run python scripts/verify_dreamina_canvas_skills.py after upgrading: the packaged Skills are pinned to an upstream snapshot, and the ledger format is versioned.

Current packaged version: 0.5.1+codex.20261018 (development preview).

Contributing and support

Open functional issues at . Read CONTRIBUTING.md and SECURITY.md first, and state the CLI version you verified against, whether the change alters approval binding or the ledger format, and include the affected tests.

License

Apache-2.0 — see LICENSE, NOTICE and THIRD_PARTY_NOTICES.md.