AI Agent Hub
Back to plugins
🧰

dsh-json-patch

Web Tools Updated 2026.08.19

Run the following command in DeepSeek Harness:

dsh plugin install uckkk/dsh-json-patch

Paste the following prompt into your AI chat to install this plugin:

Run dsh plugin install uckkk/dsh-json-patch in DeepSeek Harness to install this plugin; full source is available at https://github.com/uckkk/dsh-json-patch .

About this plugin

If you often need to make precise, incremental changes to JSON documents inside an agent conversation—and especially in settings where extra network calls or third-party services are unacceptable—dsh-json-patch fills that gap. It provides a pure Node implementation of the JSON Patch (RFC 6902) interface with zero network access and zero external service dependencies, so every patch operation stays entirely on your local machine.

The core capability is clean and focused: the plugin registers a set of RFC 6902 standard operations such as add, remove, and replace as conversation tools, letting you apply standardized, auditable edits to structured JSON data. No data ever leaves your machine, which makes this a natural fit for scenarios where local execution and data sovereignty are strict requirements.

If your workflow repeatedly involves fine-grained structural edits to config objects, API responses, or document fragments, and you want a lightweight, dependency-free, fully open-source option, dsh-json-patch is a drop-in solution. It is released under the MIT license, and the entire source is transparent so you can review every line before installing.

Use Cases

  • Applying standard JSON Patch operations to structured data inside an agent conversation
  • Editing config objects or API responses in offline or data-sovereignty-restricted environments
  • Making auditable, fine-grained incremental edits to document fragments without any network calls

Best For

  • Developers who frequently handle JSON data within LLM agent workflows
  • Engineering teams that require all operations to run locally with data never leaving the machine
  • Engineers who prefer lightweight, dependency-free, fully auditable tooling