creating-opencode-plugins
Skill124 starsChanged 8 months agoMITView on GitHub
Use when creating OpenCode plugins that hook into command, file, LSP, message, permission, server, session, todo, tool, or TUI events - provides plugin structure, event API specifications, and implementation patterns for JavaScript/TypeScript event-driven modules
- Reads credentials
We split this skill into the rules it gives an agent
2 rules. About 0 are common, 2 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 320
When porting Claude Code hook behavior to OpenCode plugins, use these event mappings:
- Distinctive: line 379
When porting Claude hooks to OpenCode plugins, you'll need to rewrite the logic using the OpenCode plugin API.
≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Creating OpenCode Plugins
Overview
OpenCode plugins are JavaScript/TypeScript modules that hook into 25+ events across the OpenCode AI assistant lifecycle. Plugins export an async function receiving context (project, client, $, directory, worktree) and return an event handler.
When to Use
Create an OpenCode plugin when:
- Intercepting file operations (prevent sharing .env files)
- Monitoring command execution (notifications, logging)
- Processing LSP diagnostics (custom error handling)
- Managing permissions (auto-approve trusted operations)
- Reacting to session lifecycle (cleanup, initialization)
- Extending tool capabilities (custom tool registration)
- Enhancing TUI interactions (custom prompts, toasts)
Don't create for:
- Simple prompt instructions (use agents instead)
- One-time scripts (use bash tools)
- Static configuration (use settings files)
Quick Reference
Plugin Structure
export const MyPlugin = async (context) => {
// context: { project, client, $, directory, worktree }
return {
event: async ({ event }) => {
// event: { type: 'event.name', data: {...} }
switch(event.type) {
case 'file.edited':
// Handle file edits
break;
case 'tool.execute.before':
// Pre-process tool execution
break;
}
}
};
};Event Categories
| Category | Events | Use Cases |
| command | command.executed | Track command history, notifications |
| file | file.edited, file.watcher.updated | File validation, auto-formatting |
| installation | installation.updated | Dependency tracking |
| lsp | lsp.client.diagnostics, lsp.updated | Custom error handling |
| message | message.*.updated/removed | Message filtering, logging |
| permission | permission.replied/updated | Permission policies |
| server | server.connected | Connection monitoring |
| session | session.created/deleted/error/idle/status/updated/compacted/diff | Session management |
| todo | todo.updated | Todo synchronization |
| tool | tool.execute.before/after | Tool interception, augmentation |
| tui | tui.prompt.append, tui.command.execute, tui.toast.show | UI customization |
Plugin Manifest (package.json or separate config)
{
"name": "env-protection",
"description": "Prevents sharing .env files",
"version": "1.0.0",
"author": "Security Team",
"plugin": {
"file": "plugin.js",
"location": "global"
},
"hooks": {
"file": ["file.edited"],
"permission": ["permission.replied"]
}
}Implementation
Complete Example: Environment File Protection
// .opencode/plugin/env-protection.js
export const EnvProtectionPlugin = async ({ project, client }) => {
const sensitivePatterns = [
/\.env$/,
/\.env\..+$/,
/credentials\.json$/,
/\.secret$/,
];
const isSensitiveFile = (filePath) => {
return sensitivePatterns.some(pattern => pattern.test(filePath));
};
return {
event: async ({ event }) => {
switch (event.type) {
case 'file.edited': {
const { path } = event.data;
if (isSensitiveFile(path)) {
console.warn(`⚠️ Sensitive file edited: ${path}`);
console.warn('This file should not be shared or committed.');
}
break;
}
case 'permission.replied': {
const { action, target, decision } = event.data;
// Block read/share operations on sensitive files
if ((action === 'read' || action === 'share') &&
isSensitiveFile(target) &&
decision === 'allow') {
console.error(`🚫 Blocked ${action} operation on sensitive file: ${target}`);
// Override permission decision
return {
override: true,
decision: 'deny',
reason: 'Sensitive file protection policy'
};
}
break;
}
}
}
};
};Example: Command Execution Notifications
// .opencode/plugin/notify.js
export const NotifyPlugin = async ({ project, $ }) => {
let commandStartTime = null;
return {
event: async ({ event }) => {
switch (event.type) {
case 'command.executed': {
const { command, args, status } = event.data;
commandStartTime = Date.now();
console.log(`▶️ Executing: ${command} ${args.join(' ')}`);
break;
}
case 'tool.execute.after': {
const { tool, duration, success } = event.data;
if (duration > 5000) {
// Notify for long-running operations
await $`osascript -e 'display notification "Completed in ${duration}ms" with title "${tool}"'`;
}
console.log(`✅ ${tool} completed in ${duration}ms`);
break;
}
}
}
};
};Example: Custom Tool Registration
// .opencode/plugin/custom-tools.js
export const CustomToolsPlugin = async ({ client }) => {
// Register custom tool on initialization
await client.registerTool({
name: 'lint',
description: 'Run linter on current file with auto-fix option',
parameters: {
type: 'object',
properties: {
fix: {
type: 'boolean',
description: 'Auto-fix issues'
}
}
},
handler: async ({ fix }) => {
const result = await $`eslint ${fix ? '--fix' : ''} .`;
return {
output: result.stdout,
errors: result.stderr
};
}
});
return {
event: async ({ event }) => {
// Monitor tool usage
if (event.type === 'tool.execute.before') {
console.log(`🔧 Tool: ${event.data.tool}`);
}
}
};
};Installation Locations
| Location | Path | Scope | Use Case |
| Global | ~/.config/opencode/plugin/ | All projects | Security policies, global utilities |
| Project | .opencode/plugin/ | Current project | Project-specific hooks, validators |
Common Mistakes
| Mistake | Why It Fails | Fix |
| Synchronous event handler | Blocks event loop | Use async handlers |
| Missing error handling | Plugin crashes on error | Wrap in try/catch |
| Heavy computation in handler | Slows down operations | Defer to background process |
| Mutating event data directly | Causes side effects | Return override object |
| Not checking event type | Handles wrong events | Use switch/case on event.type |
| Forgetting context destructuring | Missing key utilities | Destructure { project, client, $, directory, worktree } |
Event Data Structures
// File Events
interface FileEditedEvent {
type: 'file.edited';
data: {
path: string;
content: string;
timestamp: number;
};
}
// Tool Events
interface ToolExecuteBeforeEvent {
type: 'tool.execute.before';
data: {
tool: string;
args: Record<string, any>;
user: string;
};
}
interface ToolExecuteAfterEvent {
type: 'tool.execute.after';
data: {
tool: string;
duration: number;
success: boolean;
output?: any;
error?: string;
};
}
// Permission Events
interface PermissionRepliedEvent {
type: 'permission.replied';
data: {
action: 'read' | 'write' | 'execute' | 'share';
target: string;
decision: 'allow' | 'deny';
};
}Testing Plugins
// Test plugin locally before installation
import { EnvProtectionPlugin } from './env-protection.js';
const mockContext = {
project: { root: '/test/project' },
client: {},
$: async (cmd) => ({ stdout: '', stderr: '' }),
directory: '/test/project',
worktree: null
};
const plugin = await EnvProtectionPlugin(mockContext);
// Simulate event
await plugin.event({
event: {
type: 'file.edited',
data: { path: '.env', content: 'SECRET=123', timestamp: Date.now() }
}
});Real-World Impact
Security: Prevent accidental sharing of credentials (env-protection plugin blocks .env file reads)
Productivity: Auto-notify on long-running commands (notify plugin sends system notifications)
Quality: Auto-format files on save (file.edited hook runs prettier)
Monitoring: Track tool usage patterns (tool.execute hooks log analytics)
Claude Code Event Mapping
When porting Claude Code hook behavior to OpenCode plugins, use these event mappings:
| Claude Hook | OpenCode Event | Description |
PreToolUse | tool.execute.before | Run before tool execution, can block |
PostToolUse | tool.execute.after | Run after tool execution |
UserPromptSubmit | message.* events | Process user prompts |
SessionEnd | session.idle | Session completion |
Example: Claude-like Hook Behavior
export const CompatiblePlugin = async (context) => {
return {
// Equivalent to Claude's PreToolUse hook
'tool.execute.before': async (input, output) => {
if (shouldBlock(input)) {
throw new Error('Blocked by policy');
}
},
// Equivalent to Claude's PostToolUse hook
'tool.execute.after': async (result) => {
console.log(`Tool completed: ${result.tool}`);
},
// Equivalent to Claude's SessionEnd hook
event: async ({ event }) => {
if (event.type === 'session.idle') {
await cleanup();
}
}
};
};Plugin Composition
Combine multiple plugins using opencode-plugin-compose:
import { compose } from "opencode-plugin-compose";
const composedPlugin = compose([
envProtectionPlugin,
notifyPlugin,
customToolsPlugin
]);
// Runs all hooks in sequenceNon-Convertibility Note
Important: OpenCode plugins cannot be directly converted from Claude Code hooks due to fundamental differences:
- Event models differ: Claude has 4 hook events, OpenCode has 32+
- Formats differ: Claude uses executable scripts, OpenCode uses JS/TS modules
- Execution context differs: Different context objects and return value semantics
When porting Claude hooks to OpenCode plugins, you'll need to rewrite the logic using the OpenCode plugin API.
Schema Reference: packages/converters/schemas/opencode-plugin.schema.json
Documentation: https://opencode.ai/docs/plugins/
In this repository64 other files it gives its agents.
- AGENTS.mdAGENTS.md
- CLAUDE.mdCLAUDE.md
- beanstalk-deployCursor rule
- .cursor/rules/core-principles.mdcCursor rule
- .cursor/rules/creating-agents-md.mdcCursor rule
- .cursor/rules/creating-claude-agents.mdCursor rule
- .cursor/rules/creating-claude-commands.mdCursor rule
- .cursor/rules/creating-claude-hooks.mdCursor rule
- .cursor/rules/creating-continue-packages.mdCursor rule
- .cursor/rules/creating-copilot-packages.mdCursor rule
- .cursor/rules/creating-cursor-commands.mdCursor rule
- .cursor/rules/creating-cursor-rules.mdcCursor rule
- .cursor/rules/creating-kiro-agents.mdcCursor rule
- .cursor/rules/creating-kiro-packages.mdCursor rule
- .cursor/rules/creating-skills.mdcCursor rule
- .cursor/rules/creating-windsurf-packages.mdCursor rule
- .cursor/rules/elastic-beanstalk-deployment.mdCursor rule
- .cursor/rules/format-conversion.mdcCursor rule
- .cursor/rules/github-actions-testing.mdcCursor rule
- .cursor/rules/karen-repo-reviewer.mdcCursor rule
- .cursor/rules/prpm-json-best-practices.mdcCursor rule
- .cursor/rules/self-improve-cursor.mdcCursor rule
- .cursor/rules/testing-patterns.mdcCursor rule
- .cursor/rules/typescript-type-safety.mdcCursor rule
- .cursor/rules/typescript-type-specialist.mdcCursor rule
- creating-agent-skillsSkill
- adding-new-ai-formatSkill
- agent-builderSkill
- aws-beanstalk-expertSkill
- beanstalk-deploySkill
- claude-hook-writerSkill
- creating-agents-mdSkill
- creating-claude-agentsSkill
- creating-claude-commandsSkill
- creating-claude-hooksSkill
- creating-claude-rulesSkill
- creating-continue-packagesSkill
- creating-copilot-packagesSkill
- creating-cursor-commandsSkill
- creating-cursor-rules-skillSkill
- creating-kiro-agentsSkill
- creating-kiro-packagesSkill
- creating-opencode-agentsSkill
- creating-skillsSkill
- creating-windsurf-packagesSkill
- creating-zed-extensionsSkill
- documentation-standardsSkill
- elastic-beanstalk-deploymentSkill
- github-actions-testingSkill
- human-writingSkill
- integrating-stripe-webhooksSkill
- karen-repo-reviewerSkill
- npm-trusted-publishingSkill
- osgrepSkill
- postgres-migrationsSkill
- prpm-developmentSkill
- prpm-json-best-practicesSkill
- pulumi-troubleshootingSkill
- self-improvingSkill
- slash-command-builderSkill
- The first 60 shown.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

