opencode-model-announcer - GitLite

KO | EN

gitlite — search

#developer-tools #markdown #gemini #react #php #self-hosted #svg #cloud-native #security #proxy

opencode-model-announcer

(★ 34)

No description available.

File Explorer

ZIP Download

- - release-please.yml
- - test.md
- index.ts
- plugin.ts
- service.test.ts
- service.ts
.gitignore
AGENTS.md
bun.lock
CHANGELOG.md
eslint.config.js
LICENSE
mise.toml
package.json
README.md
release-please-config.json
RELEASE.md
tsconfig.build.json
tsconfig.json

# Use via CDN

jsDelivr

jsDelivr serves any public GitHub repository as a CDN with zero setup. Pick a version and a file to get a ready-to-paste link and snippet.

Version

File

Link

Example

Command Glossary

Commands referenced in this DOCs, explained below.

🔍

`npm install`

View Details ▼

Install Node packages.

npm {{[i|install]}}

Install dependencies listed in `package.json`:

npm {{[i|install]}} {{package_name}}@{{version}}

Download a specific version of a package and add it to the list of dependencies in `package.json`:

npm {{[i|install]}} {{package_name}} {{[-D|--save-dev]}}

Download the latest version of a package and add it to the list of dev dependencies in `package.json`:

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

This plugin automatically injects the current model name into the chat context. This helps the LLM know exactly which model it is (e.g., `google/gemini-3-pro-high` or `anthropic/claude-opus-4-5-thinking`) without needing to ask the user or hallucinating.

## Features

- **Automatic Announcement**: Injects a synthetic message part into every user prompt turn.
- **Context Awareness**: Seen only by the LLM, invisible in the TUI history.
- **Zero Configuration**: Just load the plugin and it works.

## Installation

```bash
npm install @ramarivera/opencode-model-announcer
```

Or add it to your `~/.config/opencode/opencode.json`:

```json
{
  "plugin": ["@ramarivera/opencode-model-announcer"]
}
```

## How it Works

The plugin hooks into `experimental.chat.messages.transform`. Right before the messages are sent to the AI provider, it:

1.  Finds the last user message.
2.  Resolves the **friendly model name** (e.g., "Claude 3.5 Sonnet") using the OpenCode SDK.
3.  Prepends a system-level announcement to the user's message:

`[SYSTEM: CURRENT_MODEL_ANNOUNCEMENT - The current model being used is: {friendlyName} ({provider}/{model}). Use this information to tailor your responses if necessary.]`

🤖 _This content was generated with AI assistance using Claude Opus 4.5._

> **Note**: Ironically, this README was generated when the project started, and I was actually using **Gemini 3 Pro**, not **Claude Opus 4.5**. With this plugin, the model would have known its true identity, and this disclaimer would have been accurate. This is exactly the problem I'm aiming to solve.

## Disclaimer

This is a community plugin and is not officially affiliated with, endorsed by, or sponsored by [OpenCode](https://opencode.ai) or [SST](https://sst.dev).