Release Notes: v0.12.0¶
Release Date: 2026-06-29
Overview¶
OmniAgent v0.12.0 introduces AI image generation via OmniImage, new CLI commands for interactive setup and diagnostics, GitHub skill integration, web content extraction skill, and auto-reply system for automated responses. This release also includes security hardening for process execution and HTTP handling.
Highlights¶
- Image Generation - OpenAI-compatible API endpoints for AI image generation via OmniImage (DALL-E, FLUX)
- GitHub Skill - Search issues, PRs, and code across GitHub repositories
- Web Content Skill - Extract and process content from URLs
- Auto-Reply System - Automatic message responses with configurable triggers
- Interactive Setup Wizard - Guided CLI setup with
omniagent setup - Doctor Command - Diagnose configuration and connectivity issues
- Tool Call Visualization - Real-time tool call display in web UI with expand/collapse
- Tool Usage Statistics - Track tool call counts and last used timestamps
- Security Hardening - Process tree termination on timeout and bounded HTTP reads
New Features¶
Image Generation¶
Generate images via OpenAI-compatible API endpoints:
# Enable image generation
export IMAGE_ENABLED=true
export IMAGE_PROVIDER=openai # or: fal
export IMAGE_MODEL=gpt-image-2
omniagent openai serve
API endpoints:
| Endpoint | Description |
|---|---|
POST /openai/v1/images/generations |
Generate images from prompt |
POST /openai/v1/images/edits |
Edit images with mask |
POST /openai/v1/images/variations |
Create image variations |
Supported providers:
- OpenAI: DALL-E 2, DALL-E 3, gpt-image-2
- Fal AI: FLUX Pro, FLUX Dev, FLUX Schnell
See Image Generation Guide for details.
GitHub Skill¶
Search and interact with GitHub repositories:
import "github.com/plexusone/omniagent/skills/compiled/github"
a, err := agent.New(config,
agent.WithCompiledSkill(github.NewSkill(os.Getenv("GITHUB_TOKEN"))),
)
Tools provided:
| Tool | Description |
|---|---|
github_search_issues |
Search issues across repositories |
github_search_prs |
Search pull requests |
github_search_code |
Search code in repositories |
github_get_issue |
Get issue details |
github_get_pr |
Get pull request details |
Web Content Skill¶
Extract content from URLs for processing:
import "github.com/plexusone/omniagent/skills/compiled/web"
a, err := agent.New(config,
agent.WithCompiledSkill(web.NewSkill()),
)
The web_fetch tool extracts text content from web pages, handling JavaScript rendering when needed.
Auto-Reply System¶
Configure automatic responses based on message patterns:
agent:
auto_reply:
enabled: true
triggers:
- pattern: "out of office"
response: "I'm currently away. I'll respond when I return."
- pattern: "urgent"
response: "This has been flagged as urgent. Notifying immediately."
See Auto-Reply Guide for configuration options.
Interactive Setup Wizard¶
New omniagent setup command for guided configuration:
The wizard helps configure:
- LLM provider and API keys
- Channel connections (WhatsApp, Telegram, Discord)
- Voice settings
- Skills and tools
Doctor Command¶
Diagnose configuration and connectivity:
Checks:
- API key validity
- Provider connectivity
- Channel status
- Skill requirements
Sessions Command¶
Manage conversation sessions:
Status Endpoint¶
New endpoint for gateway status and version information:
Returns:
{
"status": "ok",
"version": {
"version": "0.12.0",
"commit": "abc1234",
"build_date": "2026-06-29",
"go_version": "go1.26.4",
"platform": "darwin/arm64"
}
}
Tool Call Visualization¶
The web UI now displays tool calls in real-time during streaming responses:
- Expandable tool call blocks with name and status
- Formatted JSON display of tool arguments
- Status indicators: Running, Complete, Error
- Click to expand/collapse tool details
Tool Usage Statistics¶
Track and display tool usage across sessions:
# Get usage summary for all tools
curl http://localhost:8080/api/v1/tools/usage
# Get stats for a specific tool
curl http://localhost:8080/api/v1/tools/web_fetch/stats
The tools panel in the web UI displays:
- Call count per tool
- Last used timestamp
- Top tools by usage
Security Fixes¶
Process Tree Termination¶
When exec tools timeout, the entire process tree is now terminated, preventing orphaned child processes from consuming resources.
Bounded HTTP Reads¶
HTTP response bodies are now bounded to prevent out-of-memory conditions from malicious or malformed responses.
New Environment Variables¶
| Variable | Description | Default |
|---|---|---|
IMAGE_ENABLED |
Enable image generation | false |
IMAGE_PROVIDER |
Provider: openai, fal |
openai |
IMAGE_MODEL |
Default model | - |
IMAGE_API_KEY |
API key override | - |
FAL_KEY |
Fal AI API key | - |
Configuration Changes¶
New Fields¶
| Field | Type | Description |
|---|---|---|
image.enabled |
bool | Enable image generation |
image.provider |
string | Provider: openai, fal |
image.model |
string | Default model |
image.api_key |
string | API key |
Image Configuration¶
New API Endpoints¶
| Endpoint | Description |
|---|---|
GET /api/v1/status |
Gateway status with version info |
GET /api/v1/tools/usage |
Tool usage summary |
GET /api/v1/tools/{name}/stats |
Stats for specific tool |
Dependency Updates¶
| Package | Change |
|---|---|
github.com/plexusone/omniimage |
v0.1.0 (new) |
Upgrade Guide¶
From v0.11.0¶
- Update your dependency:
- (Optional) Enable image generation:
- (Optional) Add GitHub skill:
import "github.com/plexusone/omniagent/skills/compiled/github"
a, err := agent.New(config,
agent.WithCompiledSkill(github.NewSkill(token)),
)
Documentation¶
New guides added:
- Image Generation - AI image generation setup
- Web Content Skill - URL content extraction
- Auto-Reply - Automatic response configuration
Updated guides:
- OpenAI-Compatible API - Added image endpoints
- Configuration Reference - Image config
- Environment Variables - Image variables
- Memory Guide - Short-term and long-term memory
Full Changelog¶
See CHANGELOG.md for the complete list of changes.