In the DeepSeek Harness (DSH) plugin ecosystem, image generation capabilities are often scattered across different plugins. To unify access and simplify configuration, blackdm666/dsh-plugin-88api-image provides an integrated image studio. The plugin consolidates the originally separate Image2 and Nano Banana models into a single entry point, allowing users to use one key and a single installation to call six models from 88api.ai for text-to-image generation, multi-reference image editing, and high-resolution output.

Core Features

This plugin mainly addresses the problems of fragmented image generation models and cumbersome configuration. Its specific capabilities are as follows:

  1. Unified access: Merges the Image2 and Nano Banana models, routed through the API provided by 88api.ai.
  2. Generation and editing: Supports text-to-image generation and sequential editing based on multiple reference images.
  3. High resolution: Supports 2K and 4K resolution output.
  4. Batching and persistence: Supports sequential batch tasks; default model can be set and retained across sessions.
  5. Secure configuration: Provides masked key configuration and security management.

Installation and Activation

When installing, specify the target Profile. Assuming the web Profile is used:

dsh plugin --profile web add github:blackdm666/dsh-plugin-88api-image

For DSH Desktop, replace web with the active Desktop Profile name. After installation, it is recommended to verify that the configuration is active before startup:

dsh --profile web --dump-config

The output should contain a line with 88api-image.

Configure Key

The plugin supports two configuration methods. Interactive command-line input is recommended to avoid exposing the key in command history.

  1. Run the key setting command:
   dsh plugin --profile web exec 88api-image set-key
  1. Wait for the terminal to display 88API Key:.
  2. Enter the key and press Enter (input is hidden at this point).

You can also pipe the key for non-interactive configuration:

printf '%s' "$API88_KEY" | dsh plugin --profile web exec 88api-image set-key --stdin

Check the current configuration:

dsh plugin --profile web exec 88api-image config

The configuration file is stored by default in ~/.dsh/88api-image/config.json.

Set Default Model

You can set the default model via command line or natural language instructions to the Agent.

dsh plugin --profile web exec 88api-image set-default gemini-3-pro-image

The default model is gpt-image-2, but any default setting saved by the user will override it.

Using Tools

The plugin provides three primary model-facing tools:

  • image_generate_88api: Generates or edits images. Supports specifying model, count (number of images per generation), prompts (sequential batch prompts), reference_images (reference images), and dry_run (validation only, no charge).
  • image_model_88api: Lists models, reads, or changes the default model.
  • image_key_88api: Checks, saves, or clears the key.

Generated images are saved by default in ~/Pictures/88api-image. If the DSH model supports image input and an attachment service is mounted, generated images are automatically attached to the tool result.

Security and Precautions

  • Runtime: The plugin runs in the Host process and does not have a sandboxing mechanism. Therefore, users should review the source code themselves and pin the version (install by specifying a commit SHA).
  • Communication security: HTTPS communication is required.
  • Feature scope: The current version (0.1) does not support older features such as Codex-only SSE preview, folder/template workflows, repair-pass, and nail stress-test.

Summary

The plugin integrates multiple image generation models into DSH through a unified interface. It is suitable for developers who need to manage multiple image generation strategies within a single workflow. Complete documentation and source code are available on GitHub.