OB-1 built OB-1. PM to PR, around the clock. If it can scale us, it can scale the next 10 engineers in your org.
Capabilities
The system that builds software while you sleep, and learns to do it better each time.
Spawn agents that work autonomously on issues while you focus elsewhere. Connect anytime to check progress or take over.
Every PR traces back to the conversations that created it. Click any code section to see the original discussion.
Agents run in isolated environments with filesystem and network controls. Test, build, and verify code autonomously without touching your production systems.
Track the equivalent engineering hours delivered by agents. See cost per PR, turns per task, and productivity trends.
Access Claude Opus 4.5, GPT-5.2, Gemini 3 Pro, Grok 4.1, and more through a single interface. Auto-routing picks the best model for each task.
OB-1 learns from your session history to auto-configure its own expertise. Skills evolve based on what works, getting better at your specific workflows over time.
Changelog
The frontier of agentic coding, right in your terminal.
After each response, OB-1 suggests follow-up prompts based on context. Quickly continue workflows, explore alternatives, or dive deeper with one click.
Share any conversation with a link. Recipients can view your session or use --clone to fork it and continue independently with full context preserved.
read_file now handles images, audio, PDFs seamlessly. New get_video_frames tool extracts frames for visual analysis. Paste images directly into prompts.
Run /analyze to spawn 3 parallel agents that investigate friction points, workflow patterns, and codebase structure. Results surface in your inbox as actionable recommendations.
HandoffTool and ReturnTool enable orchestrated agent collaboration. Main agent delegates tasks to specialists, tracks progress, and synthesizes results.
Toggle into read-only planning mode where agents explore codebases and design approaches before execution. Get your implementation strategy approved before any code changes.
Agents can now pause and ask clarifying questions when requirements are ambiguous. Reduces wasted iterations by gathering context before committing to an approach.
One command to set up MCP integration with IntelliJ, PyCharm, WebStorm, and other JetBrains IDEs. Seamless context sharing between your IDE and terminal agent.
Visual TUI picker to browse your conversation history and jump back to any point. Restore context, fork from past decisions, or undo agent mistakes.
New reactive architecture for tool execution. Parallel tool calls are scheduled efficiently, with real-time UI updates and proper cancellation handling.
Skills system is now stable and enabled by default. Create custom skills with activate_skill, or use built-in skills for PR reviews, planning, and Graphite workflows.
A personalized inbox that surfaces actionable tips based on your usage patterns. OB1 learns how you work and proactively suggests skills, commands, and MCPs to boost your productivity.
Run /optimize to analyze your conversation history and get personalized recommendations. Discovers unused skills, suggests sub-agents for repetitive tasks, and identifies MCP integrations that match your workflow.
One-command migration from Claude Code. Imports your existing skills, agents, commands, and AGENTS.md configurations. Preserves your customizations while upgrading to OB1's enhanced capabilities.
Create checkpoints in any conversation and branch off to explore alternatives. Fork your session mid-task to try different approaches without losing your original context.
Dedicated agent for comprehensive code review. Analyzes security, correctness, test coverage, and code quality. Run automatically after completing features or manually with delegate_to_agent.
Skills are now activated on-demand with specialized guidance for specific tasks. PR reviews, planning, and workflow-specific skills inject expert knowledge exactly when needed.
Tracks concrete outcomes: files staged, commits made, PRs submitted via Graphite/GitHub CLI. Understands the difference between running git add vs git status for measuring real impact.
OB1 updates itself automatically on startup without interrupting your workflow. Shows a subtle 'updated' notification only after successful update.
Delegate complex tasks to purpose-built agents: Explore for fast read-only codebase searches, Plan for architectural design, General for multi-step modifications.
User-friendly error handling for common issues like 413 (payload too large) - explains what happened and suggests solutions instead of raw API errors.
Prevents agent from calling file tools with incorrect extensions. Automatic verification before reading or writing files.
LLM-powered session extraction that creates persistent memory across conversations. Summaries from past sessions are injected into new prompts.
Completely bypass all output height constraints for the current view. Useful for reviewing massive diffs or lengthy error stacks.
Focus into embedded shell sessions for direct interaction. The agent pauses while you type commands directly, then resumes when you're done.
Context window optimization that automatically compacts old tool results with "[Tool result compacted]" markers when approaching context limits.
Async cost tracking with generation stats, token usage, and context utilization. Warns when approaching model limits.
Paste images from clipboard or drag screenshots into the prompt. Handles volatile macOS temp screenshots that disappear after drag-and-drop.
Intelligent truncation of large tool outputs based on token budgets, keeping the most relevant content while respecting context limits.
Built-in task management with write_todos tool. Agent breaks down complex tasks, tracks progress, marks items complete. Toggle visibility with Ctrl+T.
Intelligent model routing that analyzes your prompt complexity in real-time. Routes simple tasks to fast models (Haiku) and complex tasks to premium models (Opus/Sonnet).
Name sessions with /session name <tag> for easy recall. Fork at any point with the thread tool to branch conversations. Resume with /resume.
Access and fork past conversation threads. Full thread history with search, continuation, and branching capabilities.
In-depth web research that synthesizes information across multiple sources. Handles complex questions requiring comparison and nuanced understanding.
Configurable sound alerts for task completion, errors, and attention-needed events. Toggle with /notifications command or OB1_NOTIFICATIONS env var.
When using weighted presets or Auto Mode, OB1 filters out models whose context windows are too small for your current conversation.
Configurable thinking budgets per model/agent. Sub-agents like Plan use thinkingBudget: -1 (unlimited) for deep reasoning on architectural decisions.
Fetch content from any URL including localhost and private networks. Supports up to 20 URLs per request with custom processing instructions.
Access 300+ models from OpenAI, Anthropic, Google, Meta, Mistral and more through a single unified API with automatic prompt caching detection.
Safe processing of images, PDFs, and other binary files with automatic size checking and limits. Groups files under 3MB per turn.
Intelligent output management for shell commands. Automatic use of quiet flags, temp file redirection for large outputs, and smart truncation.
Fetch content from any URL including localhost and private networks. Supports up to 20 URLs per request with custom processing instructions.
Fast, optimized code search with automatic output limiting (max 20k matches). Supports pagination, regex patterns, and file type filtering.
Precise text replacement requiring 3+ lines of context. Supports multiple replacements with exact match verification to prevent accidental changes.
Automatic analysis of existing code patterns, test frameworks, and styling conventions. Ensures all generated code matches project standards.
Persistent storage for user-related facts and preferences. Remembers coding styles, common paths, and personal tool aliases across sessions.
Efficient file matching with modification time sorting. Supports case-sensitive search and respects .gitignore patterns.
Join the waitlist for early access, or talk to us about enterprise deployment.